Firmable is the market-leading B2B sales intelligence platform in Asia Pacific — and we're scaling that success globally at pace. Backed by leading investors and 2,000+ customers, we exist to give sales teams an unfair advantage: the deepest company and people data of any platform, enriched with real-time signals, served at the right moment by intelligent agents.
The backend services, APIs, and AI systems behind every Firmable product are the engine room. This role owns it.
The Role
As Software Engineering Lead, you'll own the function that designs, builds, and operates the backend services, RESTful APIs, and AI/LLM systems powering every Firmable product. You'll lead two engineering pods through their team leads while staying deeply hands-on yourself. Reporting to the Head of Data & Services.
This is not a people-management seat. ~80% hands-on architecture and engineering, ~20% leading the function. You whiteboard the system, write the reference implementation, then raise the whole team to that bar. If you want to step away from the code, this isn't it.
What You'll Own
Backend and API architecture — Firmable's Python services (FastAPI, Pydantic, Logfire): the secure, well-documented backbone every product, internal agent, and external customer depends on
Data modelling — schemas, access patterns, and integrity at scale across PostgreSQL, Snowflake, and DynamoDB
AI/LLM services and agents — production systems that are fast, idempotent, observable, and cost-bounded; every LLM call a named, versioned tool with prompts, schemas, traces, and evals
Harness engineering — the agent runtime the team builds on: tool-calling and MCP server layers, structured-output validation, bounded agentic loops, sandboxed execution, human-in-the-loop checkpoints, replayable traces
Eval harnesses — scorers, labelled sets, trajectory and multi-turn evals, LLM-as-judge calibration, regression suites in CI, drift detection on vendor model updates
Scale and reliability — systems serving millions of API calls: idempotency, caching, rate limits, cost ceilings, circuit breakers, bounded retries, graceful degradation
Observability standard — structured traces, p50/p95/p99 by stage, cost per request, and the dashboards that make it legible
Function leadership — lead and grow two pods through their leads; run architecture and code reviews; set the bar for quality, testing, and AI-native practice
Cross-functional glue — connect Data Platform, Application, and Product; turn early ideas into architecture decisions and working prototypes fast
What We're Looking For
Must Haves
7-10 years hands-on backend engineering, including time leading engineers or engineering leads with 3+ years of hands on experience with Python.
Built and operated large-scale production systems — millions of API calls, and you know what breaks first when load grows
Expert Python with deep production experience in FastAPI
Shipped AI/LLM services, APIs, and agents in production — and stood up the evals that keep them trustworthy
Harness engineering experience — you've built the scaffolding agents run inside: tool-calling frameworks, MCP servers, eval and test harnesses other engineers depend on, and you can walk us through it in detail
Hands-on data modelling across SQL and NoSQL
Designed and consumed RESTful services at scale
Shipped real work with agentic IDEs — Claude Code, Cursor, or equivalent. Not "tried it" — your default mode
Lead from the front — whiteboard to reference implementation to a team operating at that standard
Highly Valued
Eval frameworks (Braintrust, Inspect AI, Promptfoo) and tracing stacks (Logfire, OpenTelemetry)
Agent frameworks and protocols — Claude Agent SDK, Pydantic AI, LangGraph, MCP
AWS and/or GCP at scale — ECS/Fargate, Lambda, S3, RDS, DynamoDB
B2B data, entity resolution, or high-throughput enrichment pipelines
Snowflake and modern data-platform tooling
Startup or scaleup experience where you shipped fast and owned outcomes end to end
How We Build
Firmable is an AI-native organisation. Agentic development, AI-driven test generation, automated review pipelines, and built-in evals and traces are the default mode of working, not a productivity experiment. Recurring workflows ship as versioned SKILL.md specs any teammate or agent can run.
We run lean and ship fast — small senior teams, no layers, minimal process, weekly releases moving toward daily. Teams own their stack end to end. You need to be working this way already, and able to bring a team with you.
Why This Role
Own the engine room — you're not joining a team, you're leading the function
Architectural latitude — the next generation of our services and AI platform is yours to shape
Hands-on leadership — the influence of a lead with the satisfaction of still building
Massive leverage — your work reaches every customer, every day
Competitive base + meaningful equity — a share in the upside we're building toward
is an equal opportunity employer. We believe diverse teams build better products.
Originally posted on