Senior Software Engineer - Prompt Engineering

Description :

- Design & build production services and UIs, from APIs to frontends, with an eye for simplicity, performance, and operability.

- Own systems instrument, monitor, and debug in prod; drive incident response and post-mortems; automate everything.

- Evolve our AI stack prototype and productionize features that depend on LLMs - prompt pipelines, evals, guardrails, retrieval, and memory.

- Tune & measure experiment with model selection,finetuning, quantization, LoRA,PEFT, structured outputs, function,tool calling, and latency,cost tradeoffs.

- Ship great APIs design pragmatic REST, GraphQL, and,or gRPC interfaces; versioning, pagination, authN,Z, schema evolution, and backward compatibility.

- Collaborate tightly with product,design to turn ambiguous problems into lovable, secure, and compliant experiences.

- Mentor teammates via reviews, tech talks, and crisp docs.

What makes you a great fit :

- Curiosity as a habit you read release notes, try new runtimes and models, and build weekend prototypes for fun, a lot of your Youtube consumption is hacking and learning new technologies, and you enjoy learning what you dont know on your own

- Self-driven you find the problem behind the problem, align stakeholders, and land outcomes without hand-holding.

- Systems mindset you think in SLOs, budgets (latency,cost,error), blast radius, and graceful degradation.

- API ergonomics you care about naming, error design, rate limits, observability, generated SDKs, and docs that dont lie.

- Data foundations you know when to pick Postgres vs columnar vs KV vs vector; you`ve shipped schema migrations and zero-downtime deploys.

- LLM-practical you`ve built RAG or agents in anger; you understand context windows, tokenization, evals, and prompt,tooling hygiene.

- Security instincts you default to least privilege, tame secrets, and design auth flows that survive real traffic.

Nice to have :

- Hands-on with vector DBs (e. FAISS, HNSW, Milvus, PGVector), embeddings pipelines, and hybrid ranking.

- Can Design retrieval layers build embeddings pipelines, vector indexes, hybrid search (BM25 + ANN), chunking,merging strategies, and memory graphs.

- Protocol-fluent you actually enjoy HTTP,1.1 vs HTTP,2,3 quirks, caching semantics, content negotiation, and idempotency.

- Experience with model hosting (self-hosted inference servers, serverless GPUs, or managed endpoints) and caching,streaming strategies.

- Infra chops containers, IaC, CI,CD, service meshes, feature flags, canary,blue-green, cost observability.

- Frontend XP with modern TypeScript frameworks and component systems; accessibility and performance budgets.

- Prior startup experience or meaningful open-source contributions