How Work Actually Gets Done
At Sourcegraph, decisions flow through a public handbook and written proposals rather than meetings or managerial decree. Every major call (product direction, architectural choices, policy changes) lands as a request for comments (RFC) in the company handbook, the single source of truth. This isn't process for show; it reflects a belief that whoever is closest to the code or customer should steer the ship, and that clarity comes from writing, not hallway conversations lost in Slack threads.
The company went fully remote on February 20, 2020, and never looked back. Today it employs 165 people across more than 30 countries, with engineering split roughly evenly between U.S. and EU time zones, plus a smaller slice elsewhere. That spread shapes the daily rhythm: no one expects real-time responses. Teammates hand off work in writing, document progress in shared docs, and wait for feedback cycles that stretch across days, not minutes. Meetings are treated as exceptions, not defaults. The handbook spells out norms for communication, code review, and collaboration, reinforcing a culture where thoroughness beats speed.
But this system carries weight. The async model rewards discipline and self-management, yet demands constant documentation. Engineers don't just write code — they write proposals, update handbooks, and explain trade-offs in prose anyone can read. Some employees describe the load as heavy, even exhausting. The handbook has grown thick enough that newcomers spend weeks just catching up. And while transparency lives in every layer (code, issues, RFCs) the volume of written norms can slow decisions when urgency clashes with process.
The pace accelerated sharply after 2021. Under CEO Quinn Slack, Sourcegraph shifted from a self-hosted enterprise model to shipping Cody, its AI coding assistant, and core code search features multiple times per day. That change followed a deliberate pivot led by new head of engineering Steve Yagi, who joined in 2022 after coming out of retirement. Yagi brought structure to chaos: faster builds, tighter feedback loops, and a focus on speed without sacrificing quality. Where customer feedback once took a month to surface, it now flows in hours.
Still, not everyone adapts. Former employees note that the intensity of the pace (combined with frequent reorganizations and strategic shifts) can feel destabilizing. Layoffs in 2023 and 2025, along with a late-2025 strategic split that carved out certain product lines, introduced uncertainty into what had been a stable remote culture. For some, the promise of autonomy turned into ambiguity about direction.
Yet the core operating principle holds: decisions belong to those doing the work. Whether an engineer proposing a new API design via RFC or a designer pushing back on a product timeline in a public thread, Sourcegraph leans into decentralized judgment. It works well for self-starters and deep thinkers — and poorly for those who need tight oversight or clear top-down mandates.
The remote-first setup means no corner office, no watercooler politics, and no informal mentorship over coffee. Instead, recognition happens through written contributions, code commits, and documented impact. For many, that feels fair. For others, it feels isolating.
Values and Operating Principles
Sourcegraph's operating principles trace back to its founding premise: software development had become too manual, and the company set out to fix that with tools developers would actually use every day. CEO Quinn Slack frames this as a product-model fit problem, not a long-term differentiation play. "You cannot predict what the world is going to be like in a year," he said in a 2025 interview. "So pursuing some long-term differentiation strategy is folly." That thinking shapes how the company evaluates features, prioritizes engineering work, and structures decisions across a fully remote workforce.
The public employee handbook documents core processes and serves as the primary coordination mechanism for a team spread across time zones. Sourcegraph adopted a model similar to GitLab: some code is open source, some is not. The open-source portions live under the Apache 2.0 license, a choice Slack defended in a 2021 interview. "We didn't feel like copyleft was the right choice for us." The enterprise version includes telemetry for update checks, but the self-hosted open-source version does not. That split reflects a broader principle: features that matter more to large organizations than individual developers belong in the enterprise tier.
This distinction drives product decisions. Sourcegraph's customers include every developer at Stripe, Palo Alto Networks, four of the six largest U.S. banks, and seven of the ten biggest tech companies. Trust with those customers, Slack told Contrary Research, is the most valuable asset. "That customer trust is valuable. But everything else, you can't really plan for." It also enables choices that competitors call "crazy," like building Amp as a multi-tenant agent with no model-selection dropdown. While rivals dig into privacy, self-hosting, and cost arguments, Sourcegraph uses customer trust to build what it calls the best agentic coding tool.
The async, writing-heavy culture reinforces this approach. Built In notes that Sourcegraph's "defining tradeoff is a handbook-first, async, writing-heavy culture: autonomy and transparency are high, but so is the documentation load and reliance on written debate over meetings." Employees receive a generous travel budget to meet teammates globally, and the company hosts Merge twice a year as an onsite alignment event. Decisions happen through written proposals and debate rather than meetings, favoring strong writers comfortable operating without constant synchronous input.
Speed and user retention take precedence over launch-day metrics. Slack told Contrary Research: "Do not look for that launch-day pop. Look for the thing that will get users to keep using and loving it, that will have a huge impact." That philosophy applies internally as well. Sourcegraph moves fast enough to pivot when model capabilities shift, but Slack warns against overfitting to what worked at a given moment. Windsurf's "Legacy mode" exemplifies the trap he wants to avoid: "That's really overfitting to what worked at a given period of time, and not setting yourselves up to evolve with your customers and with the model capabilities."
Open source plays a supporting role rather than a leading one. Early on, Sourcegraph released its code hoping community contributions would follow. "Any contributions that we would get from the community would be a pure bonus," Slack said in a 2021 interview. The reality, he admitted, was underwhelming: "Open sourcing something, people don't care unless they care about that product." That lesson shaped a product strategy focused on utility over ideology. Code search, Slack argued, is something developers "won't be able to live without once you have it." The same logic applies to Amp and Cody: build tools people depend on daily, and trust follows.
The culture rewards curiosity and directness. Employees are expected to own outcomes, write clearly, and ship incrementally. At the same time, the documentation burden and async-first norms can slow decisions for people who prefer synchronous collaboration. Transparency extends to compensation, which Built In San Francisco describes as a strength alongside async operating norms and high ownership. Restructuring-driven change fatigue and isolated integrity concerns appear in employee feedback as counterpoints.
One tension stands out: Sourcegraph pushes hard on engineering-led decisions and radical transparency, yet it also operates with tight strategic discipline around customer trust and product focus. The company's public values emphasize openness, but its competitive strategy hinges on moves that competitors view as counterintuitive. That contradiction isn't accidental. It reflects a culture that treats principles as hypotheses to test, not doctrine to defend.
What the Hiring Bar Selects For
Sourcegraph's hiring bar centers on three non-negotiable traits: intellectual honesty, deep technical substance, and written-first communication. Everything else is negotiable.
The technical screen sets the tone early. Candidates face an online assessment or take-home exercise that rewards timed accuracy and clear explanation of approach — not clever one-off tricks. This means demonstrating the ability to ship working code under pressure while articulating reasoning, a pattern that recurs throughout the loop.
Interviews then probe fundamentals, system or domain depth, and how candidates recover from ambiguity. This isn't trivia — it's whether you can walk through a compiler architecture or explain why indexing depth matters beyond "grep at scale." Candidates with compiler or language-server backgrounds, or production AI-coding experience, have clear edges because they can speak to the real engineering tradeoffs Sourcegraph builds on daily.
Code review friendliness weighs as much as correctness. The same source states that "candidates are evaluated on code-review friendliness as much as correctness." This means writing code that others can read, accept, and extend — a signal that maps directly to Sourcegraph's handbook-first, async culture where most decisions happen without meetings.
Remote-work effectiveness gets probed explicitly and early. TechInterview.org puts it plainly: "remote-work effectiveness is probed early — the company hires remote-only and needs candidates who genuinely thrive async." Weak async communication practices from past roles are a red flag: candidates who reflexively reach for "a quick call" to resolve everything signal poor fit for a written-first culture where decisions happen in documents, GitHub issues, and chat threads.
Domain knowledge matters. Not having used Cody or competitor AI coding tools seriously before interviewing raises red flags. So does treating code search as "just grep at scale" — the indexing depth is real engineering. And no opinions about the AI-coding landscape won't cut it; Sourcegraph operates in that space daily, and candidates must form their own current opinion based on recent usage.
The hiring-manager and final panel rounds calibrate level and team fit. ClavePrep advises bringing impact metrics, a conflict/recovery story, and a crisp "Why Sourcegraph?" that mentions the business unit — not just the brand. Generic answers fail here.
Common funnel mistakes reveal what the bar actually selects against. Skipping the OA or skills gate because "I'll wing the interview" backfires. Memorizing Sourcegraph trivia instead of rehearsing structured problem-solving misses the point. Sending one generic résumé to every Software posting under the Sourcegraph brand shows no understanding of the role differentiation. And surprising HR late with location, notice, or authorization constraints signals poor project management — a trait that doesn't translate well to a company that treats the hiring process as a project.
Green flags in the process confirm these priorities. ClavePrep identifies clear stage lists from recruiters, predictable follow-ups, written confirmation of next steps, and interviewers who have read your résumé as signs of process health. Yellow flags — long silence after a strong round without a stated SLA, last-minute panel changes, conflicting descriptions of the role level — suggest internal disarray but not necessarily rejection.
The length itself is a filter. Glassdoor reports the process often exceeds eight steps, and ClavePrep notes U.S.-facing loops span 2–6 weeks for mid-level roles, with senior loops running longer when panels and compensation approvals stack. Ghost periods after a strong round usually mean internal scheduling, not automatic rejection — but candidates should nudge once politely.
Candidates who treat the Sourcegraph hiring process as a project (with artifacts, timelines, and rehearsal) outperform those who only wait for HR to email. This isn't accidental; it mirrors how Sourcegraph expects work to get done once you're inside.
One tension in the research: while the bar emphasizes substance over pedigree, the process length and documentation demands can filter out qualified candidates who lack the time or bandwidth to invest heavily in a multi-week loop. The company's 3.3/5 culture rating on Glassdoor, 11% lower than the IT industry average, suggests some candidates find the bar's execution uneven even when the stated principles remain clear.
What Current and Former Employees Say
Public reviews paint a company split between its engineering culture and its leadership instability. On Blind, Sourcegraph holds a 3.6 out of 5 stars rating from 36 company reviews as of 2026, with employees praising compensation and benefits while criticizing management. SimplyHired's 2026 figures show higher marks (4.7 for pay and benefits, 3.7 for work-life balance) but management still trails at 3.3. The contrast reflects a pattern that runs through nearly four years of employee commentary: strong technical culture undermined by weak executive stewardship.
The criticism centers on leadership. Multiple reviews from 2024 describe a CEO who "thinks he knows best and doesn't take input, from anyone," with data and facts failing to sway direction. One February 2024 reviewer wrote that "big decisions with wide ranging repercussions are done in a vacuum," while another that same month accused the C-suite of painting "a rosy picture, when often it is so far from reality." Turnover in upper management has been "unprecedented," according to a October 2022 post, sending "negative working relationships with top leadership" rippling through the organization.
Layoffs amplified the distrust. A July 2022 review described cuts "mostly to upper management," but a November 2024 post accused leadership of "using the possibility of being laid off to drive individuals to only follow orders and never speak up." The same reviewer called the number of layoffs "surprising for a company of this size." That anxiety persists: an October 2024 comment noted that "the thought of losing your job is forefront on everyone's mind."
Strategy confusion compounds the instability. Reviews from 2022 through 2024 repeatedly flag "constant change in company strategy and direction," "product strategy is a bit scattered," and "lack of a long term strategy and vision." An October 2022 post worried that "product strategy doesn't seem to be working since sales pipeline is stalling," while a December 2022 reviewer warned that "ARR has plateaued and the strategy for reaching out to strategic customers...has to pay off or the company is essentially in trouble." Revenue figures from GetLatka show growth ($10M in December 2021, $31M in September 2024, $50M in March 2025) but the pace of strategic pivots has left teams uncertain about priorities.
The human cost shows in attrition data scattered across reviews. "Top tier talent is leaving in droves," wrote one February 2024 reviewer. Others described "high attrition this year" (December 2022), "lots of turnover lately and top talent leaving" (October 2022), and "constant turnover of executives and management" (September 2022). A December 2022 post captured the mood: "feels like the internal atmosphere has changed this year, people are on edge."
Despite the turbulence, a consistent counter-current of praise appears across reviews. Engineering teams drew repeated compliments: "Product and Engineering team are world class" (December 2021), and one 2022 reviewer called it "a company that would be able to endure an even prolonged economic downturn." Remote-first benefits resonated — multiple posts highlighted "location independent pay," "30 mandatory minimum PTO days," "forced async culture," and "good async culture." Technical growth opportunities earned mentions, along with perks like "insurance is fully covered" and "great company trips."
The compensation picture is mixed. SimplyHired's 2026 data shows software engineers at $178,333 and engineering managers at $217,500, with Blind rating compensation at 4.0 out of 5. But some reviewers warned that pay "is good but not great" (January 2022) or "not very competitive, especially given the inflation of the last year" (December 2021). A March 2025 post advised early-career engineers to "runaway to public companies get the RSU."
| Role | Source | Salary |
|---|---|---|
| Software Engineer | SimplyHired (2026) | $178,333 |
| Engineering Manager | SimplyHired (2026) | $217,500 |
Cultural strengths persist even as leadership falters. Reviews consistently praised "smart people to work with," "great colleagues," and "the product is awesome and you get to eat your own dog food as a developer." One 2022 reviewer noted "culture amongst the company is top notch, everyone communicates well and is all in on transparency." The open-source ethos drew positive mentions, as did Sourcegraph's position tackling "Big Code" problems.
The tension between these two realities (strong engineering culture versus weak executive leadership) defined the employee experience. Sourcegraph's remote-first, engineering-led model delivered on technical collaboration and async work, but the company's leadership instability and strategic churn had eroded confidence. As one November 2024 reviewer concluded, "it would behoove them to listen to constructive criticism but at this point it's a bit late given the competition in the space has surpassed them."
Who Thrives Here and Who Burns Out
Sourcegraph's all-remote, engineering-led setup favors people who can operate with minimal direction across time zones. The company employs 165 people globally, with engineering split roughly evenly between U.S. and EU time zones and a small slice elsewhere, according to a 2023 profile in The Pragmatic Engineer newsletter. That structure rewards engineers who can write clear documentation, ship asynchronously, and make decisions without waiting for consensus. It penalizes people who need frequent check-ins or thrive on in-person collaboration.
The "transparent by default" principle extends internally, making information far more accessible to employees than at most organizations. That openness suits candidates who prefer to dig into raw data rather than navigate hierarchy. It also means work lands in public view quickly, which fits engineers comfortable defending their choices in writing. Those who expect private feedback loops or gradual visibility ramps often find the pace jarring.
Sourcegraph pays globally equal salaries, a practice that attracts talent in lower-cost regions but can create friction for employees in expensive markets who feel undercompensated relative to local peers. The company's Glassdoor culture rating sat at 3.3 out of 5 as of September 2026, 11% below the IT industry average, suggesting the trade-offs are real. Reviews flagged strengths in transparent pay and async norms alongside challenges from restructuring-driven change fatigue and heavy documentation demands.
Engineers who enjoy building developer tools for other developers tend to thrive. Sourcegraph's core products (Code Search and Cody, the AI coding assistant) serve technical audiences, so the day-to-day work appeals to people who find satisfaction in solving complex problems for their peers. The company's customer base includes Uber, Plaid, and four of five FAANG companies, which adds stakes but also credibility for contributors who want their work used at scale.
However, the same transparency that empowers some employees can exhaust others. A former team member described feeling constantly scrutinized, with every decision documented and visible across the organization. That intensity suits people who process feedback quickly and don't personalize criticism. It burns out those who need space to iterate privately before sharing ideas.
The company's recent alliance with Leidos to bring AI-enabled software development tools to government customers adds another layer. The partnership targets sensitive missions, which raises the bar for code quality and security. Engineers who thrive under high-stakes conditions (where a bug could affect national security systems) may find the work energizing. Those who prefer lower-pressure environments often struggle with the weight of that responsibility.
Recent job postings show Sourcegraph is still hiring aggressively for remote roles, including an ML & Agentic Systems Engineer and a Tech Lead for Code Plane. Those positions signal the company's focus on advancing its AI capabilities, which means engineers comfortable with rapid iteration and emerging tooling will find opportunities. Those who prefer stable, well-defined processes may find the pace of change destabilizing.
The tension between autonomy and intensity defines the culture. People who value ownership and trust tend to excel. Those who need structure and predictability often burn out.
Working in AI? Zero G Talent tracks the openings: see every open Sourcegraph role, browse AI jobs, the companies hiring, and the people building the field.