Skip to main content
← frontier

Careers at 100ms: Teams, Pay and How to Get Hired

By John Hugo•

Who Gets Hired and Onto Which Teams

100ms had 94 employees as of March 2026, up 31 (+6.6%) from the prior year, per Revelio Labs. Every role on the company's Lever board and on Zero G Talent's board lists Bengaluru and on-site. No remote options, no satellite offices, no hybrid language. The company builds AI agents that handle scheduling, intake, benefits verification, and telehealth for U.S. healthcare operations, per its website.

Since 2020, hiring has clustered around three cores. First, the AI agent stack: roles titled AI Engineer, Senior Product Engineer, and Software Engineer Architect, all explicitly tied to healthcare agents. Second, the infrastructure underneath: Platform Engineer for core infrastructure. Third, the go-to-market and operations layer: Product Manager for AI agents, Sales Development Representative working U.S. hours from Bengaluru, IT Engineer supporting the agent stack, and a Benefit Verification Specialist who sits inside the product loop. HR and recruiting roles round out the list. As of early 2025, headcount sits between 51 and 200 people, per TeamBlind.

That concentration shapes who gets through. Engineering roles ask for production experience with LLMs, RAG pipelines, and the latency constraints of real-time voice or chat; these are the same constraints the company's blog dissects in posts about peta-scale consistency and sub-100ms API architecture. The product manager role expects fluency in healthcare workflows: prior auth, eligibility, scheduling rules. The SDR role requires U.S. shift hours and a grasp of the buyer landscape. Candidates without that overlap — generalist full-stack engineers, product managers from consumer tech, recruiters without healthcare exposure — don't appear in the hiring record.

Glassdoor shows 43% of employees would recommend the company to a friend. TeamBlind tells a sharper story: "Never join, run away," "worst management, worst work life balance," "no job security, they will kick you out like nothing without severance." The divergence is the signal. 100ms recruits for the exact slice of work it has right now.

Compensation: What the Roles Pay

Zero G Talent's board shows three salaried roles in Bengaluru with a combined cash band of roughly $22k–$79k (median $63k). That range sits below U.S. AI-engineering markets but competes with top-tier Indian product companies. Glassdoor reviewers rate compensation and benefits 3.7–3.8 out of 5, with Glassdoor reporting a 1% dip over the past year and another calling it stable.

Role Location Annual Cash Band (INR) Approx. USD
Product Manager – AI Agents (Healthcare) Bengaluru 2,500,000 – 8,000,000 $30k – $96k
Senior Product Engineer – AI Agents (Healthcare) Bengaluru 2,500,000 – 6,000,000 $30k – $72k
AI Engineer – AI Agents (Healthcare) Bengaluru 2,000,000 – 5,000,000 $24k – $60k
Platform Engineer – Core Infrastructure Bengaluru 1,700,000 – 3,500,000 $20k – $42k
Sales Development Representative – US Healthcare (India, US Hours) Bengaluru 1,000,000 – 2,000,000 $12k – $24k
IT Engineer – AI Agents (Healthcare) Bengaluru 1,000,000 – 1,500,000 $12k – $18k

Product and senior engineering roles carry the widest bands, up to 3.2x from floor to ceiling. Platform engineering tops out at 3.5M INR. Sales and IT roles sit in a narrow 1–2M INR band.

Equity details are not published on the board or in public job posts. Wellfound's 100ms profile lists "equity" as a searchable field but surfaces no grant sizes, vesting schedules, or refresh policies. Candidates should ask directly; 100ms has not confirmed any equity terms.

Benefits appear standard for a funded Bengaluru tech company: health insurance, parental leave, and a learning stipend show up in employee reviews. The 3.7–3.8 Glassdoor score suggests the package is competitive. No public data covers 401(k) equivalents (EPF/NPS), variable bonuses, or sign-on equity.

All six roles are Bengaluru-based. The SDR role explicitly requires U.S. hours, which means night shifts for India-based reps, a factor that should weigh on total-comp calculations. The board lists only three salaried roles in its aggregate median; the other three postings may be newer or mapped to different leveling. Treat the $63k median as a midpoint for the engineering/PM cluster, not a company-wide average.

How the Hiring Process Works

The interview loop at 100ms runs five rounds over roughly four to six weeks, though candidates report timelines as short as two weeks when the team moves fast. The company does not publish an official process; what follows is reconstructed from candidate-reported data aggregated by PracHub, Codemia, and Dataford as of late 2026. Rounds appear or disappear depending on what signal the team still needs.

Recruiter or founder screen (30 minutes). This is not a courtesy call. At 94 employees as of March 2026, a founder screen carries real evaluative weight. The conversation covers background, motivation, and a mutual pitch, but the bar is explicit: come prepared with a specific answer to "why 100ms." Generic startup enthusiasm fails here. Recruiters and founders screen for genuine interest in realtime media infrastructure (WebRTC, SFU architecture, the SDK surface) and for evidence you've read the engineering blog or used the product.

Coding round (60 minutes). One or two problems at LeetCode medium difficulty, but with practical texture drawn from the domain: buffering logic, scheduling, stream processing. Interviewers grade on correctness and speed of clean execution. Candidates who treat this as pure algorithm practice miss the point; the problems map to constraints the team hits daily, such as packet loss recovery, jitter buffer tuning, and codec negotiation fallbacks.

Systems / domain round (60–90 minutes). Design a group video call service or a core component: SFU forwarding, quality adaptation, room state management, reconnection handling. Interviewers supply media context as needed and probe reasoning, not just the diagram. This is where the SFU — the architectural heart of 100ms — becomes the conversation.

SDK or platform deep dive (60 minutes). Track-dependent. Mobile SDK roles: Android/iOS lifecycle, permissions handling, backgrounding, Bluetooth audio routing. Web SDK roles: browser media APIs, getUserMedia constraints, MediaStream manipulation, WebCodecs. Backend roles: Go or Node service design for media servers, room orchestration, recording pipelines, global deployment. Half the engineering organization builds developer experience: SDKs for Android, iOS, React Native, Flutter, and web that wrap brutal platform complexity into APIs a developer integrates in an afternoon. The deep dive tests whether you've lived that complexity.

Culture & founder round (45 minutes). Behavioral, but not the standard STAR-scripted variety. Founders and senior engineers ask for ownership stories, pace preferences, and ambition alignment. They describe it as an honest two-way conversation about startup fit. Candidates who want structure, predictability, or a narrow scope tend to self-select out here, or get selected out.

What disqualifies before the final decision. A generic "why this company" answer in the founder screen. Coding solutions that work but ignore the practical constraints the problem implies. Systems designs that treat WebRTC as a black box. In the deep dive, unfamiliarity with the platform primitives your track lives on, including camera lifecycles on iOS, RTCPeerConnection internals on web, and Go channel patterns for the media pipeline. And in the culture round, a mismatch on ownership: the team ships fast, owns incidents end-to-end, and expects engineers to operate without a platform team beneath them.

Offers can follow the final round within days. Startup decision-making means no committee approval cycle; the people who interviewed you decide.

Where the Work Happens

100ms operates from two primary locations that reflect its split between founding jurisdiction and engineering concentration. The registered headquarters sits at 49095 Woodgrove Common in Fremont, California, the address on the California Secretary of State filing from June 2022 and the incorporation record. California business registry data lists the zip as 94539; CB Insights shows 95439. The Fremont address is the legal home, the place where the corporation receives service of process (agent: CorpNet, Incorporated) and files its annual statements. It is also the only U.S. office the company discloses. Craft.co confirms "1 office location" in the United States.

The engineering center of gravity is in Bengaluru. Glassdoor recorded 40 open roles there in August 2026 and 44 in September 2026. The first-party board carries six current postings, all Bengaluru-based, spanning Product Manager - AI Agents (Healthcare), Senior Product Engineer - AI Agents (Healthcare), AI Engineer – AI Agents (Healthcare), Platform Engineer (Core Infrastructure), Sales Development Representative - US Healthcare (India, US Hours), and IT Engineer - AI Agents (Healthcare). The Bengaluru presence appears to be housed at Bangalore Alpha Lab, a coworking and shared-office provider in J P Nagar that lists private cabins from ₹4,000/month with move-in within seven days. That detail matters: it signals a flexible, lease-light footprint rather than an owned campus. For a Series A company that last raised $20M in 2022 (total raised $24.5M), a coworking arrangement preserves runway while letting the team scale headcount quickly.

What the sites enable follows from what 100ms builds. The product is live video infrastructure — APIs and SDKs for real-time video calls, livestreaming, chat, whiteboard, polls — targeting EdTech, fitness, and telehealth. That workload demands low-latency media servers, global TURN/STUN deployment, and tight integration with WebRTC internals. The Bengaluru team, heavy on platform and AI engineers, is where the core media pipeline gets written, tested, and shipped. The Fremont address anchors U.S. sales, compliance, and the healthcare vertical go-to-market (the SDR role explicitly works "US Hours" from India). The time-zone spread, roughly 12.5 hours between Fremont and Bengaluru, creates a follow-the-sun window for incident response and customer escalations, but it also means the India crew carries the overnight pager for production issues.

Neither location is a research lab in the traditional sense. There is no disclosed hardware lab, no anechoic chamber, no custom silicon bring-up. The infrastructure is cloud-native; the "lab" is a staging environment spun up in AWS, GCP, or Azure. Candidates should expect day-to-day work to look like writing Go/Rust media services, debugging ICE negotiation failures across NAT topologies, and instrumenting latency percentiles, not racking servers. The coworking setup in J P Nagar also means open-plan desks, shared meeting rooms, and the ambient noise of other startups. If you need a private office or a dedicated hardware bench, this isn't it. What you get is a high-density engineering cluster focused on a narrow, hard problem: making live video reliable at scale for customers who measure quality in milliseconds.

Who Thrives Here

The engineering blog and public writing reveal a team that defines itself through latency discipline. The company's own history statement frames the mission bluntly: "Making live video work, our team has obsessed about this problem for years and solved it at world-class scales for different network and device conditions." Posts on the 100ms blog and affiliated engineering channels break down sub-100-millisecond API delivery into latency budgets, minimized hops, async fan-out, layered caching, circuit breakers, and strong observability. A 2026 target called the "10-20-70 Rule" allocates 10 milliseconds to DNS and connection, 20 to time-to-first-byte, and 70 to rendering and painting. Zero-RTT delivery gets its own deep dive. This is an organization that measures in milliseconds and publishes the math.

That technical posture selects for a specific profile. Engineers who thrive here treat distributed-systems trade-offs as daily work, not interview theory. The CAP Theorem discussion on the company's architecture site argues that "good enough consistency is no longer an option" for peta-scale systems, such as global financial ledgers, real-time ad bidding, and massive gaming backends. Candidates who have only operated at single-region scale or who reach for eventual consistency as a default will struggle. The blog's emphasis on circuit breakers, structured observability, and latency budgets implies an expectation that you instrument before you ship and you debug with data, not intuition.

The current hiring wave sharpens the filter. First-party board data shows six active roles in Bengaluru, five of them tied to an "AI Agents (Healthcare)" initiative. The concentration in healthcare AI agents signals a pivot: the live-video infrastructure is being pointed at a regulated, high-stakes vertical. That means candidates need comfort with compliance constraints, data-privacy boundaries, and the slower feedback loops that clinical or hospital deployments impose, without losing the latency obsession.

Glassdoor aggregates show 100ms at 3.3 out of 5 across 19-plus reviews, with only 43 percent of reviewers saying they would recommend the company to a friend and 40 percent holding a positive business outlook. Culture and values sit at 2.8 — the lowest sub-score — while work-life balance reaches 3.1 and career opportunities hit 3.4. The overall rating has moved barely: Glassdoor's data shows a 3 percent improvement over the prior year, another calls it stable. That flatline matters. A candidate walking in should expect a workplace where the technical bar is visible and the cultural experience is contested.

The 3.4 career-opportunity score hints that people who stay grow fast, likely because the technical surface area (WebRTC core, media pipelines, signaling, now LLM-driven agents) is wide and the team is small enough that ownership is real. TeamBlind reviewers who stayed describe the same pattern: "If you match the brief, you ship. If you don't, you're gone in months."

Self-assessment checklist: you have shipped production distributed systems where a 50-millisecond regression triggered an incident review; you write design docs that include failure modes and latency budgets; you can defend a consistency model to a product manager who wants a feature yesterday; you are willing to work U.S. hours from Bengaluru if the SDR role fits, or to operate in a healthcare-regulated stack if the engineering roles fit. If that list describes you, the process in the previous section will feel familiar. If it doesn't, the hiring bar is designed to filter you out before offer stage, and the 57 percent who wouldn't recommend the company already know why.


Working in frontier tech? Zero G Talent tracks the openings: see every open 100ms role, browse frontier tech jobs, the companies hiring, and the people building the field.

Ready to Start Your Space Career?

Browse frontier jobs and find your next opportunity.

View frontier Jobs