Founding Hacker

Read this twice We build in the deep end with sensitive health data. We're selective by design. A performance regression doesn't just slow things down; it can delay care. High agency, low ego: insight → plan → ship, fast. Relentless execution: compress years into months without losing judgment or quality.

Builder operator: You ship it and you run it. SLOs, cost, performance, and reliability.

Build, break, learn, and ship across core product, internal platforms, and high upside experiments.

Own end-to-end: data models, security posture, UX, reliability, performance, and cost.

Instrument everything: metrics, traces, logs, and audits wired into CI/CD and SLO gates.

Optimise hot paths: shrink p99s, remove 30 to 50% latency, and cut waste without adding complexity.

Lead incidents well: contain, communicate, recover, write the postmortem, and fix the root cause.

Shape the team: define standards, run tight interviews, build a culture of taste and speed.

5+ years shipping production systems at real scale (1M+ DAU, 10k+ RPS, or TB per day pipelines) with end-to-end ownership.

~ Regulated environments: shipped in healthcare, fintech, or payments; implemented HIPAA, ABDM, PCI, or RBI controls, data residency, and passed audits.

~ Security first engineering: threat modelling, mTLS, key rotation, tokenisation, least privilege IAM; you spot data leak risks in reviews.

~ SRE discipline: define SLIs and SLOs, lead incidents, write postmortems that drive fixes, design for graceful degradation.

~ Range across the stack: backend, infra, and modern frontend; you can own a feature from RFC to prod and take on a call.

~ Clear writing: docs and diagrams that non-engineers and auditors can understand.

RLS, tenant isolation, tokenisation or deterministic encryption, retention and deletion workflows, immutable audit trails.

OSS impact: meaningful PRs or maintainer roles in PostHog, Prometheus, Grafana, OpenTelemetry, ClickHouse, Kafka or Redpanda, Envoy, Temporal, Kubernetes, Sentry, Qdrant or pgvector.

Published depth: papers, conference talks, or a serious blog series on systems, performance, security, or data platforms.

India regulatory exposure: you've navigated audits and localisation.

TEEs such as Nitro Enclaves, differential privacy, k-anonymity or l diversity, or practical ZK or MPC exposure.

ML in production: retrieval pipelines and eval harnesses, redaction and guardrails, serving with vLLM or TGI on GPUs, Hugging Face Hub for models and datasets.

Performance and migrations: flamegraphs and eBPF, lock contention fixes, cache design, zero-downtime schema and service cutovers.

Languages and runtimes: Go, Rust, TypeScript or Node, Python; Data and streaming: SQLite; GraphQL where it helps; Frontend: React with Server Components; TypeScript; Playwright for E2E.

AWS (EKS, RDS or Aurora, S3, KMS, IAM); Cloudflare (Workers, Queues, Durable Objects); OPA or Gatekeeper or Cedar policies; tamper-evident audit logs with S3 Object Lock and Rekor; Intelligence and ML: dbt; Great Expectations for data quality; Hugging Face Hub, Transformers, Datasets, and Evaluate; Langfuse for LLM tracing and evals; hybrid retrieval with BM25 plus vectors via pgvector or Qdrant.

DX and quality: Tilt or Skaffold for local K8S; We ship weekly, often daily, and measure impact in p99, error budgets, RPO and RTO, and cost per 1k requests.

Ownership is end-to-end: RFC → design → code → tests → rollout → on call → postmortem → fix.

You optimise for work-life balance over growth, impact, and ownership.

You can't explain technical decisions clearly to non-engineers and auditors.

You avoid writing: specs, diagrams, and postmortems that lead to change.

You resist data: you won't set SLIs or SLOs, measure cost, or run experiments.

You think innovation, speed, performance, and reliability are someone else's job.

You write to think, debate respectfully, and change your mind with new data.

You get energy from building, operating, and iterating with users in the loop.

DM with 2 to 3 links to work you're proud of (repo, PR, X profile, talk, blog), or apply here on LinkedIn.

---