Artificial Intelligence Engineer

About GrowQR

GrowQR is an AI-native career platform. Job seekers get a team of AI coaches for resumes, interviews, roleplays, career pathways, courses and job search, unified by a conversational super-agent and a single readiness score. We serve individual seekers in India and partner with colleges, with international expansion planned.

The role

You will join a small engineering team and own features from idea to production, across the whole stack. Our backend is a mesh of Python FastAPI services coordinated by an orchestrator over Redis Streams, using Google ADK, A2A and MCP. The product runs on the latest cutting-edge models, including real-time voice and video sessions, and ships on AWS. The work is broad: new coaching features, scoring and assessment engines, cross-agent memory, real-time voice and video, billing and plan gating, admin and institution tooling, and the frontend surfaces that expose them. A growing part of it is building production-ready autonomous agentic systems: agents that plan, use tools, run on schedules, learn from feedback and act within approval gates. Everything you build must fit our existing Python and FastAPI stack and conventions. This is essential. We work from written plans, verify every change in the browser before it ships, and land every fix with the test that would have caught it.

Responsibilities

• Design, build and ship product features end to end: data model, service code, orchestrator routing, prompts, evaluations and the frontend surface.

• Build autonomous agent capabilities: multi-step planning, tool use, memory, scheduled runs, approval-gated actions and clear audit trails.

• Own LLM output quality through evals, golden sets, calibration and shadow runs before any prompt or model change ships.

• Extend multi-agent orchestration, cross-agent memory and retrieval with embeddings.

• Harden real-time voice and video sessions for latency, reconnection, session limits and cost.

• Manage the model layer as a product surface: evaluating new models, migrations, vendor drift, token budgets and caching.

• Debug production issues using Sentry, PostHog and service logs, write root-cause analyses and land fixes with tests.

• Deploy to production and QA environments on AWS and keep CI green across repositories.

Requirements

• 5+ years of production Python, with strong FastAPI experience.

You must be able to work natively within our existing FastAPI codebase and conventions rather than introduce a parallel stack.

• Shipped LLM-powered features to real users, including prompt iteration, evaluation and cost control.

• Built and operated agentic systems in production: tool calling, multi-agent orchestration, structured output, retrieval, streaming, guardrails and failure handling.

• Working knowledge of Redis, PostgreSQL, Docker and a major cloud provider, and comfort reading container logs and slow queries.

• Strong testing discipline, including test-first development.

• Fluency with an AI coding harness such as Claude Code, and the judgement to direct it well.

• Clear written communication: plans, RCAs and runbooks.

Preferred

• Google ADK, A2A or MCP.

• Real-time audio or video with live multimodal model APIs, LiveKit or WebRTC.

• DSPy or another prompt-optimisation framework.

• Next.js and TypeScript, enough to ship your own UI.

• Scraping and data pipelines (Apify).

• Privacy engineering: consent, data erasure, DPDP or GDPR.

Tech stack

Python 3.12, FastAPI, SQLAlchemy, Alembic, DSPy, latest frontier LLMs and real-time multimodal APIs, Google ADK, Redis, PostgreSQL with pgvector, Docker, AWS (EC2, RDS, ALB, WAF, SES, S3, CloudFront), nginx, Next.js 16, React 19, TypeScript, Tailwind, Clerk, Sentry, PostHog, Claude Code.