Staff Backend Engineer (remote-friendly)

Hamming is hiring a Staff Backend Engineer to own the infrastructure that 10x-100x's the platform's scale over the next several months — and to architect what comes next. The team is at the inflection where most customers do ~1,000 concurrent simulated calls, one customer needs 100K concurrent, and the next level is a million concurrent — two orders of magnitude in months. Every observability company eventually becomes a database company, and Hamming's transition is happening now. The 90-day target is to prototype Hamming's own indexing mechanism and start building toward a purpose-built database that supports multiple applications on top.

The role reports directly to Sumanyu, who personally runs engineering, product, and a chunk of sales. The team is eight people, mostly senior, ships to prod multiple times a day, and ranks #1 on Weave. The hire works alongside engineers who have shipped real things in production — a first-year Waterloo undergrad who benchmarks 99th percentile and ranks #2 on the team is the kind of teammate to expect. Sumanyu's hiring philosophy is unscalable by design: "we shape the role to the person" — strengths matter more than well-roundedness. Spike high on something specific (databases, distributed systems, observability infra, audio pipelines, low-latency systems) and the rest of the role gets built around it.

Culture is what the founders describe as Tesla-style brute-force execution — minus the code quality problems. Manic intensity, decisive operating cadence, "very low procrastination culture," "we'd rather be a little bit wrong and tweak it than theorize for multiple days." Async-default with US-hours overlap. Ships to prod many times a day. Not for chill vibes; explicitly great for the right person and explicitly wrong for most.

What You'll Own

• The simulation engine: orchestrates thousands of simulated calls, captures traces, scores them
• The red-teaming infrastructure: adversarial agents that break customers' voice agents on purpose (jailbreaks, policy violations, PII leaks) — the backend that makes these attacks repeatable and ranks them by severity
• Production monitoring: normalize ingestion of live voice-agent calls (recording, transcript, metadata), run customer assertion libraries and LLM judges across every call, surface regressions and novel failure modes before customers complain
• Customer-facing APIs and dashboards backing the evals product
• Database performance work as call volume scales 100x, and the prototype/build of Hamming's own indexing mechanism
• Ramp: month 1 = audit and fully understand the system end-to-end and start proposing solutions (crawl-walk). Month 3 = prototype the internal indexing mechanism. Month 6 = a working database / indexing layer that the platform is being built on top of

Requirements

• 5+ years of strong production backend in Python or TypeScript/Node
• Has owned real-time or near-real-time systems at scale — queues, streaming, low-latency APIs
• Has shipped end-to-end: schema design → API → observability → on-call
• Comfortable in small teams, no PM hand-holding, defines own scope
• Sets the bar without needing a title; picks the boring, durable solution when it's right; can walk a peer through a design tradeoff in a one-pager

Nice-to-haves

• First or second engineer somewhere, or one of the first ~10 hires
• Built something significant that's still running in production
• Voice, telephony, WebRTC, or audio pipeline experience
• Observability tooling, eval infrastructure, or testing platform background
• Open-source contributions to relevant stacks (LiveKit, Pipecat, Redis, Kafka, Postgres, ClickHouse, Temporal, vector DBs, etc.)
• Has worked at an early-stage startup before, not just FAANG
• Scala, Haskell, or "obscure-language mastery" background — pattern Sumanyu has seen correlate strongly with how the team approaches problems (not because Hamming uses Scala, but because the training transfers)

Trade-off flexibility

• Voice/telephony domain experience is bonus, not required — Hamming can teach the domain to a strong systems engineer faster than it can teach systems to a domain expert
• Title/seniority is flexible — output and judgment matter more than years
• Open to a great IC who doesn't want to manage, or someone ready to lead a small team within 6-12 months
• Not flexible on: ownership instinct, ability to ship without scaffolding, written communication

Location and Work Model

• Remote-friendly with US-hours overlap; preference for Austin or SF in-person
• Quarterly offsite; office in Austin (SF office coming in 2-3 months — currently co-working at investor's office in Hayes Valley)
• International candidates fine if hours overlap meaningfully and they can crush

Who Will Thrive Here

• Has a thesis about why voice AI specifically matters as a category, not just "any AI job"
• Wants to build a database for fun on weekends and would actually do it — the "obsessive towards mastery" archetype
• Power user of AI tools (Claude Code, Cursor, Codex) — has opinions about which to use when
• Played with at least one of LiveKit, Pipecat, Twilio, OpenAI Realtime, even casually
• Comfortable with extreme intensity and short time horizons — Sumanyu's framing: this is a Formula One car you only get to drive a few times in your career
• Bias toward action over whiteboard perfectionism — would rather ship and iterate than debate
• Wants to spike high on a specific superpower rather than be well-rounded — Hamming will build the role around the spike