Who Gets Hired
Sourcegraph's engineering interview funnel rejects candidates who treat code search as "just grep at scale" — a phrase interviewers hear often and treat as a disqualifier. The company builds tools engineers use to understand code: code search, navigation, batch refactoring, and the Cody AI coding assistant. The product is the filter. The company hires people who can reason about compilers, language servers, ASTs, and LLM context retrieval because that is the daily work. The question isn't whether you can code; it's whether you can operate inside the domain.
Sourcegraph's hiring process filters for specific engineering depth and collaborative traits. Candidates who prepare to demonstrate both move through; those who rely on generic applications do not. Roles exist across engineering, ML, sales, and customer success, but the bar is consistent — demonstrate you've solved problems that look like Sourcegraph's problems, and show you can do it without a shared office.
Engineering splits across code intelligence (search, navigation, symbol resolution), platform infrastructure, Cody AI, enterprise product, and developer-experience tooling. Go dominates the backend; TypeScript runs the client; Python has grown for ML and agentic systems. A Tech Lead for the Code Plane sits at IC5. An ML & Agentic Systems Engineer sits at IC4. Both are remote, like every role. Sourcegraph has been fully distributed since Quinn Slack and Beyang Liu founded it in 2013, with roughly 350 employees as of 2026 after a post-2022 rightsizing that narrowed focus back to core code intelligence plus Cody.
The company publishes its engineering handbook publicly, modeled on GitLab's approach. Candidates who read it before interviewing signal they understand the operating system they're trying to join. Remote-work effectiveness is probed in the first conversation. Interviewers ask for concrete mechanics: overlap hours you hold across time zones, decisions you moved into writing, how you unblocked yourself without waiting on a synchronous reply. Generic "I'm self-motivated" answers land flat. Candidates with language-server or compiler background have an edge; strong generalists close the gap with focused prep on Tree-sitter, LSP, and AST fundamentals. Authentic engagement with AI coding tools — having used Copilot, Cody, Cursor, or Windsurf deeply — separates candidates in the product round. Those without real usage fall into generic takes.
What the Numbers Show
Sourcegraph publishes compensation bands for every role and level: a practice the company says "creates clear accountability for compensation decisions and helps us maintain pay parity." That transparency means candidates can see the structure before they apply, but the numbers still require context.
The framework uses individual-contributor levels (IC1 through IC5) that map to scope, not tenure. Recent postings on the Zero G Talent board show the range: a Commercial Account Executive at IC2, a Customer Success Manager at IC2, an ML & Agentic Systems Engineer at IC4, and both a Tech Lead - Code Plane and an Enterprise Account Executive at IC5. Software Engineer at IC4 represents the highest total compensation reported publicly — about $278,000 including base, equity, and any bonus, per Levels.fyi data. Third-party aggregates list broader bands of $80,000–$200,000 and $130,000–$225,000 for various roles, though those figures blend roles and levels and predate the most recent hiring wave.
| Role (Level) | Reported Total Compensation |
|---|---|
| Software Engineer (IC4) | ~$278,000 (Levels.fyi) |
| Commercial AE (IC2) | Range not publicly reported for this role |
| Customer Success Manager (IC2) | Range not publicly reported for this role |
| ML & Agentic Systems Engineer (IC4) | Range not publicly reported for this role |
| Tech Lead - Code Plane (IC5) | Range not publicly reported for this role |
| Enterprise AE (IC5) | Range not publicly reported for this role |
Equity is granted at every level. The company describes its approach as "above-market" and ties grants to published bands rather than negotiating case by case. For engineering roles at IC4 and IC5, the equity component typically closes the gap between base salary and the top of the total-compensation range. Sales roles carry a different structure: variable pay depends on commission and quota attainment, and "mixed attainment figures leading to variable realized pay" mean two reps at the same level can take home materially different totals.
Benefits are generous where they apply, but explicitly limited by geography. U.S., U.K., and Canada-based employees receive 100 percent company-paid medical, dental, and vision premiums, a reproductive stipend, a health-and-fitness stipend, a home-office or coworking stipend, an education stipend, and paid parental leave. The 401(k) is offered without a match disclosed in public materials. Outside those three countries, the flagship benefits drop off; the company notes that "details differ in other locations," creating a two-tier experience for a remote-first workforce.
Time off follows a similar pattern. The policy is unlimited PTO with a documented 30-day minimum — a floor designed to "normalize longer breaks and reduce burnout by signaling leadership support for truly restorative time away." In practice, utilization "can depend on team norms and workload," and employees report the minimum is respected on some teams and treated as aspirational on others.
For candidates evaluating an offer, the levers are clear: level determines the band, location determines the benefits tier, and role type (engineering vs. sales) determines how much of total compensation is guaranteed versus variable. The published bands remove the negotiation opacity common elsewhere, but they also mean there is less room to move; preparation here means understanding which IC level matches your scope, not rehearsing a counter-offer script.
Inside the Interview Funnel
Sourcegraph's interview funnel runs five stages: application screen, an online assessment or take-home, technical interviews, a hiring-manager calibration, and offer with background checks. The company publishes this structure openly, which lets candidates prepare, but the content inside each stage filters for a specific profile. Difficulty sits at 7 out of 10 per TechInterview.org: below a Google L5 on pure algorithms, but the code-intelligence specialty and AI-coding depth create a meaningful barrier. Remote-work effectiveness is a genuine filter; candidates who haven't worked fully distributed sometimes struggle with the culture fit.
The recruiter screen lasts 30 minutes and covers background, why Sourcegraph, and team preference. A technical phone screen follows at 60 minutes: one coding problem, medium-hard. Go for backend roles, TypeScript for client, Python for ML. Problems tilt applied: implement a tree-walk for a parsed AST, handle a streaming search, implement context-ranking logic.
Many senior and staff roles add a take-home: 4 to 6 hours on a realistic engineering problem. Historically these involve code-intelligence primitives: build a small language-aware search, implement a symbol-resolution system for a tiny language, or extend a Cody-style context retrieval pipeline. The virtual onsite runs 4 to 5 rounds. Coding appears in 1 to 2 rounds: one algorithms, one applied. The applied round often involves code-intelligence primitives: AST traversal, trie-based identifier search, language-server-like symbol lookup, or context-window management for LLM prompts. System design gets one round with code-intelligence-flavored prompts: design the code-search index for 50,000 repositories with billions of symbols, design Cody's context-retrieval pipeline producing 8,000 tokens of relevant code within 500 milliseconds, or design the batch-change system that applies code modifications across thousands of repositories.
A domain deep-dive round is distinctive to Sourcegraph. Interviewers discuss language servers, AST-based search, LSIF and SCIP code-intelligence formats, symbol resolution across languages, and the specific challenges of code-at-scale. A product and craft round tests engagement with developer-experience and AI-coding-tool quality. Candidates are expected to have opinions about Cody, Copilot, Cursor, Windsurf, and the broader AI-coding space. The behavioral and hiring-manager round covers past projects, remote-work practices, and async collaboration effectiveness.
Timelines differ by track. US-facing loops for mid-level roles often span two to six weeks; senior loops can run longer when panels and compensation approvals stack. Campus and early-career funnels compress to days or roughly two weeks with OA-heavy steps and compressed interview days. Lateral and mid-level candidates see two to six-plus weeks with recruiter screen plus skills loop and deeper ownership stories. Senior and leadership tracks stretch further with multi-panel strategy and org-impact narratives and fewer timed OAs.
Recruiters screen for eligibility and genuine interest in the problem space. A strong application maps JD keywords to specific stories, not one generic résumé sent to every Software posting under the Sourcegraph brand. Candidates who treat every Sourcegraph req as the same process across campus, lateral, and leadership tracks stall early. Skipping the OA or skills gate because "I'll wing the interview" is a common funnel mistake. Surprising HR late with location, notice, or authorization constraints also derails momentum.
What disqualifies candidates: weak async communication practices in past roles — a reflex to "hop on a quick call" for everything reads as a poor fit for a written-first culture where most decisions happen without a meeting. No opinions about the AI-coding space; Sourcegraph operates in it daily. Treating code search as "just grep at scale" signals a misunderstanding of the indexing depth. Not having used Cody or competitor AI coding tools seriously before interviewing. Memorizing Sourcegraph trivia instead of rehearsing structured problem-solving. Large-employer interviews punish unstructured silence more than imperfect answers.
Green flags for candidates evaluating the process: clear stage list from the recruiter, predictable follow-ups, written confirmation of next steps, interviewers who have read your résumé. Yellow flags: long silence after a strong round without a stated SLA, last-minute panel changes, conflicting descriptions of the role level. Red flags for your decision-making: unpaid "trial work" that looks like free consulting, pressure to resign before a written offer, or refusal to state whether the req is still funded.
Interviewers listen for fundamentals, debugging habits, and how you handle follow-ups. Narrate trade-offs. When stuck, propose a simpler correct solution before optimizing. Strong answers follow patterns: ownership framed as "I owned X; partners handled Y" not "We decided…"; structure that clarifies, outlines, then details; judgment that states trade-offs and why you chose; learning that names a failure, fix, and prevention; fit for Sourcegraph that cites business-unit or software reasons, not generic brand praise. The Handbook is real — read it before interviewing. Remote work is not "remote-friendly," it's the company's structural default with no offices anywhere.
No Office, No Problem?
Sourcegraph has been all-remote since its founding in 2013. The company maintains a legal headquarters at 548 Market Street in San Francisco — a mailing address that satisfies incorporation requirements — but no operational office where employees gather. Highperformr's profile states this directly: "As an all-remote company since its inception, Sourcegraph has a truly global presence. Employees (teammates) are located in numerous countries across various time zones." The same source notes the HQ "is not a traditional operational headquarters with on-site staff" and "does not maintain a traditional physical headquarters with notable architectural features."
That description contradicts several secondary sources. Salestools.io lists "8 offices across 1 countries" and describes the San Francisco location as a "main operational hub housing key departments including engineering, product, sales, and executive leadership" with "state-of-the-art facilities" including "open-plan offices," "fitness centers," "cafeterias," and "recreation areas." Graphiq's August 2026 dossier also reports 8 total offices. These appear to be template-driven profiles that have not caught up to Sourcegraph's actual model. The first-party signal is clearer: every recent posting on the Zero G Talent board (Enterprise Account Executive, Commercial Account Executive, Sales Development Representative Team Lead, Tech Lead - Code Plane, ML & Agentic Systems Engineer, Customer Success Manager) carries the "(Remote)" designation.
The distributed model shapes how work actually happens. Highperformr characterizes the culture as "built on principles of transparency, asynchronous communication, autonomy, and high agency" with a focus on "impact and results." All functions (engineering, product development, sales, marketing, customer success) operate across time zones without a central physical anchor. That means a candidate in Berlin, Toronto, or Sydney joins the same hiring pipeline and the same daily rhythm as someone in the Bay Area. The legal HQ exists for registered-agent service and compliance; it does not confer proximity advantage.
For applicants, this removes a common filtering layer. There is no "local candidate" preference to decode, no relocation package to negotiate, no hybrid schedule to interpret. The interview process runs entirely over video and async channels, and the onboarding that follows assumes remote fluency from day one. Candidates who have only worked in office-centric environments should be ready to demonstrate how they communicate without hallway conversations, how they structure deep work without enforced quiet zones, and how they build rapport across latency.
The trade-off is intentional. Sourcegraph's minimal physical footprint reflects a bet that global talent acquisition and operational flexibility outweigh the collaboration benefits of a shared space. That bet shows up in the hiring bar: the company screens for engineers and operators who already function effectively in distributed teams. If your strongest work has happened in a room with a whiteboard, the burden is on you to prove the same output emerges when the whiteboard is a shared document and the room is a time-zone spreadsheet.
Who Stays and Why
The profile of who succeeds at Sourcegraph has shifted sharply since the company's early years. A Glassdoor review from June 2022 captures the high-water mark: "I've been in the industry for a while and haven't come across a workplace like this one. Quality and high agency are valued. I have a lot of autonomy, feel heard and I get to work with extremely smart people." That same reviewer noted the CEO and CTO are both software engineers building a developer-focused product, a dynamic that let engineers "eat your own dog food as a developer" and work on "hard challenges, in a great stack."
By late 2022 the tone fractures. A July 2022 review still praises "culture amongst the company is top notch" and "everyone communicates well and is all in on transparency." But September 2022 surfaces "constant change in company strategy and direction," "lack of focus," and "founders that lack collaboration, empathy, and people-management skills." December 2022 adds "the Open and Transparent value has given way to back channel deal making" and "management has gone from the standard issue founders-turned-C-levels problems into actively hostile." The attrition signals are consistent: "incredible amount of turnover," "high attrition this year," "lots of turnover lately and top talent leaving."
From 2023 onward the consensus hardens. February 2024: "CEO thinks he knows best and doesn't take input, from anyone… There is ZERO DEI/Diversity at the company. Most minorities have left due to the culture." June 2024: "Executive leadership swaps back and forth on strategies a lot." October 2024: "Upper management is in shambles… The new mantra is that we are a startup so expect to work nights and weekends again, otherwise we wont win." November 2024: "The leadership has turned the company completely toxic and uses the possibility of being laid off to drive individuals to only follow orders and never speak up."
Against that backdrop, the traits that currently correlate with staying and advancing read differently than the company's public messaging. The hiring process (recruiter screen, hiring-manager conversation, hands-on technical exercise tied to code search or Cody, values alignment) still selects for written communication and practical coding over live whiteboarding. But employees who describe success now emphasize tolerance for ambiguity, ability to execute without clear direction, and willingness to work extended hours. "You need to be ok with frequent changes and shifts in priorities," a June 2022 review noted presciently. "This is not a company you can get comfortable in. You're going to be challenged, constantly. And if you don't deliver, you likely won't stay around for long."
Team assignment matters more than company-wide culture. A January 2022 review puts it bluntly: "pick your team wisely… if you're not on a darling team that owns core stable features, be prepared to grind or get PIPed. I have seen collaborators burn out and leave for greener pastures within a year. Some projects have death marched for years. Work life balance isn't a company value." Conversely, engineers on favored teams report "relaxed schedules and minimal meetings."
Backgrounds that map well: engineers who have worked on developer tools, code intelligence, or large-scale code search; practitioners comfortable with Go, TypeScript, and the code-graph problem space; people who have operated in fully remote, async-heavy organizations. The board's live postings (ML & Agentic Systems Engineer [IC4], Tech Lead - Code Plane [IC5]) signal demand for depth in LLM integration and core platform architecture, not generalist web development.
The company's own careers page still reads: "We hire the most talented people in the world and give them the support they need to do their best work while thriving outside of it, too." The employee record since mid-2022 suggests the second half of that sentence no longer holds for a significant portion of the org. Candidates who join should calibrate expectations to the 2024-2025 reality: high technical bar, high volatility, team-dependent workload, and a leadership layer that multiple reviewers describe as unstable and top-down. The engineers who stay are the ones who treat that volatility as the job — not a bug to be fixed.
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.