Senior Python / AI Engineer

About Zumlo

Zumlo is a digital well-being companion that helps people understand themselves better and take small, meaningful steps toward balance every day.

It’s not another wellness app — it’s an AI-informed ecosystem that personalizes activities, reflections, and guidance across mental, emotional, physical, and spiritual well-being.

We’re post-MVP, backed by the Founder Institute, supported by a clinical advisory board, and led by a founder with 20+ years in technology and product innovation.

Our goal: build the operating system for well-being — a living, learning system that evolves with every person it supports.

The Role

We’re hiring a Senior AI Engineer who sits at the intersection of GenAI and Data Science. You’ll design grounded, human-centered AI features, build and operate the Python services behind them, and also do the modeling work: curating datasets, fine-tuning/adapting models, and running a reliable evaluation loop. You’ll balance safety, latency, cost, and outcomes, and partner closely with Mobile (React Native), Backend/Platform, Product/Design, and Clinical.

You will

• Product AI (end-to-end): ship helpful experiences across chat/coaching, summaries/explanations, guided steps, and “what to do next” — grounded with retrieval and transparent provenance.
• Retrieval & knowledge: design practical RAG pipelines (chunking strategies, embeddings, vector search, re-ranking), with freshness, dedupe, and source quality policies.
• Modeling & GenAI: curate and label datasets; run prompt-tuning / SFT / LoRA-style adaptation; experiment with embeddings and small task-specific models; export/run models efficiently (quantization/ONNX where useful).
• Evaluation (offline & online): build golden sets and adversarial suites; run shadow/canary tests; define success metrics tied to user outcomes; close the loop weekly with live signals.
• Python services & MLOps: build FastAPI-style inference and orchestration services; background jobs/queues; retries/idempotency; versioning of prompts/models/datasets; simple model registry and rollout/rollback.
• Safety & privacy by design: redaction, prompt-injection defenses, schema/guardrails, rate limits, audit trails; clear “do-not-answer” domains and escalation triggers.
• Data for decisions: trustworthy event tracking, basic cohorting, and per-feature cost/latency dashboards to guide iteration.
• Explore & de-risk: compare models/embeddings/inference options (APIs and open-weights), GPU usage when needed; prove value with small, cheap spikes before broad changes.
• Collaborate & document: work closely with Product, Mobile, Platform, and Clinical; write crisp docs/runbooks; help interview and mentor the next AI/Backend hire.

What we’re looking for

• Experience: 5+ years in software/ML engineering, including 4+ years with Python building production services or ML systems.
• Shipped GenAI features: you’ve delivered LLM + retrieval (or similar) to production and can walk through retrieval, orchestration, evaluation, and impact.
• Modeling comfort: data curation, prompt/response evaluation, small fine-tunes/adapters, embedding selection, and practical feature engineering when classical ML is the right tool.
• Backend depth: FastAPI/Django/Flask, SQL (Postgres or equivalent), background jobs/queues, CI/CD, testing discipline.
• Evaluation mindset: offline metrics that predict online behavior; knows lift vs. noise; ships, measures, and decides.
• Privacy & safety: least-privilege, secrets hygiene, safe logging; comfort working around sensitive data.
• Mindset: builder energy, clear communication, low-ego collaboration; you turn fuzzy needs into simple, reliable systems.

Preferred background (a plus, not a gate): Tier-1 universities (IIT/NIT/BITS/IIIT) or equivalent; exposure to top global schools (Ivy/Oxbridge/CMU/etc.) and high-performing teams.

Nice to have

• Azure experience (runtime, storage, monitoring) and/or Azure DevOps pipelines.
• Telemetry for AI (prompt/version tracking, cost/latency dashboards, drift checks).
• GPU/inference experience (serving/throughput) or streaming UI integrations with mobile.
• Health/well-being context or HIPAA-aware practices.

Why Join

You’ll shape how people actually experience AI in everyday well-being - grounded, helpful, and human. It’s meaningful work with room for craft, care, and impact.

Compensation & Work Model (India)

• Competitive for the India market. Additional equity or performance-based rewards may be considered based on impact and long-term fit.
• Work model: Remote to start, with required periodic in-person days in Ahmedabad; role is expected to transition to hybrid/in-office in Ahmedabad as the team scales.

To Apply

DM the founder (Harsh Sutaria) or apply here.