Principal AI Platform Engineer - Python & Generative AI

Andela; India; Hybrid; INR 3,500,000 - 6,000,000; Full time

Job summary

Andela is hiring a senior platform engineer to lead the architectural direction for an enterprise AI platform. You will prototype components, review code, and fix production systems while collaborating with multiple teams in a hands-on role.

You will own API design, security, and reliability, build reusable SDKs, and set standards for LLM/RAG workloads. The role begins remote within India for 12 months, then hybrid in Pune or Chennai.

Qualifications

• 12+ years of professional software engineering experience.
• 8+ years building and maintaining production Python services.
• Experience designing and owning web APIs with OpenAPI.
• Hands-on with LangGraph/LangChain is a plus.

Responsibilities

• Own end-to-end target-state architecture for the AI platform and its SDKs.
• Develop secure, scalable Python services and Web APIs with observability.
• Mentor senior/lead engineers through architecture reviews and hands-on guidance.
• Implement production patterns for LLM/RAG workflows with guardrails.

Skills

Python

Distributed systems

OpenAPI

Code review

Tools

Azure

AWS

Kubernetes

Docker

FastAPI

Job description

12-month fixed-term contract, temporary-to-permanent | Full-time, 40 hours per week

Remote within India for the first 12 months, then hybrid from Pune or Chennai

Approximately 6 hours of daily overlap with US Central Time is required

ABOUT THIS ROLE

Andela is hiring, on behalf of a confidential enterprise client, the most senior individual-contributor engineer on its platform engineering team. You will set the technical direction for an AI platform that other engineering teams build on: production-grade Python services and Web APIs, reusable SDKs with real downstream adoption, and the patterns that make LLM, RAG and agentic workloads safe, evaluable and repeatable at enterprise scale.

This is a hands-on Principal/Staff-calibre engineering role. You will move between target-state architecture and production code in the same week: prototyping the hard components, reviewing code, diagnosing live incidents and unblocking senior engineers. It is not an Engineering Manager, Head of AI, or diagram-only architecture role, and it carries no direct reports. Influence here comes from technical judgement and credibility, not headcount.

ENGAGEMENT, LOCATION AND WORKING PATTERN

• 12-month fixed-term contract, full-time, 40 hours per week.
• Temporary-to-permanent: the intention is to convert the successful engineer into a permanent position after the initial 12 months, subject to performance and business requirements.
• First 12 months: fully remote from within India.
• After permanent conversion: hybrid from Pune or Chennai, approximately 8 office days per month. You need to be able to work from one of those two cities long term.
• Time-zone overlap: approximately 6 hours of every working day overlapping with US Central Time. This is a genuine requirement, not a nice-to-have. The platform's stakeholders and consuming teams sit in that zone, so please only apply if you can sustain this pattern long term.
• You must be legally authorised to work in India.

WHAT YOU WILL OWN

• End-to-end target-state architecture of the AI platform: technology selection, integration patterns, migration path, technical standards and long-term evolution.
• Secure, resilient Python services and Web APIs: contract-first design, authentication and authorisation, rate limiting, performance and observability.
• Reusable Python SDKs and platform components consumed by multiple engineering teams: design, build, test, package, version and publish, with explicit semantic versioning, backwards-compatibility and deprecation policies.
• Production patterns for LLMs, RAG, agentic workflows and tool/model integration, including evaluation harnesses, guardrails, prompt practices and responsible-AI controls.
• Cloud-native delivery on Azure and/or AWS using containers, Kubernetes, serverless and infrastructure-as-code.
• Engineering quality standards: unit, integration, contract and load testing, pytest, coverage, CI quality gates, security checks and release reliability.
• Instrumentation and operability: logging, metrics, tracing, SLIs/SLOs, incident review and systemic fixes.
• The standard for AI-assisted engineering with tools such as Claude Code, Cursor or GitHub Copilot: guardrails, review practice, and measured productivity gains rolled out to other engineers.
• Mentoring of senior and lead engineers through architecture reviews, design critique and hands-on technical guidance.

WHAT WE NEED TO SEE

• 12+ years of professional software engineering, ideally with 8+ years hands-on building and maintaining production Python.
• Principal/Staff-level scope: you have owned the architecture of a platform, service family, SDK or technical domain used by multiple engineering teams.
• 6+ years building and operating production Web APIs in Python using FastAPI, Django and/or Flask, including OpenAPI and contract-first design, plus the distributed-systems thinking that goes with them.
• 4+ years of genuine SDK lifecycle ownership: packaging, semantic versioning, internal or public distribution, backwards compatibility, deprecation and breaking-change management.
• 3+ years of Generative AI in production: LLM integration, RAG, agentic workflows, and model/response evaluation and guardrails you helped design rather than merely consumed.
• Hands-on production use of LangGraph, LangChain, Semantic Kernel, MCP or equivalent frameworks.
• 4+ years deploying to Azure and/or AWS with containers, Kubernetes, serverless and infrastructure-as-code. Azure is strongly preferred.
• 4+ years of CI/CD and automated testing: Azure DevOps and/or GitHub Actions, pytest and modern quality tooling.
• Strong security and reliability fundamentals: OAuth2/OIDC, JWT, secrets management, observability, tracing, SLIs/SLOs and incident management.
• Strong data-layer depth across SQL, NoSQL and vector retrieval.
• Regular, deliberate use of AI coding assistants such as Claude Code, Cursor or GitHub Copilot, with concrete examples of measurable improvement and the guardrails you put around them.
• The communication range to explain architectural trade-offs, risk, cost and long-term consequences to both senior engineers and business stakeholders.

TECHNICAL DEPTH THAT STANDS OUT

• Agentic and multi-agent orchestration in production, not in demos.
• LLM evaluation and observability: prompt regression, LLM-as-a-judge, tracing, hallucination controls, cost optimisation and latency engineering.
• Docker and Kubernetes at scale, plus internal developer-platform or SDK ownership with meaningful downstream adoption.
• AI gateway, model routing, tool-calling governance or identity-aware agent access: the work that makes AI reusable and safe across a large organisation.
• Open-source contributions, patents, technical writing or conference speaking.

BACKGROUND WE FIND PARTICULARLY ATTRACTIVE

• Product engineering at Big Tech or comparable high-scale global engineering organisations, for example Google, Amazon, Microsoft, Meta, Apple, Netflix, Nvidia, Uber, Salesforce, Adobe or Atlassian.
• A strong Computer Science or Engineering foundation. IIT backgrounds are strongly preferred, and NIT, BITS and IIIT are also of high interest. These are prioritisation signals only, not requirements: exceptional engineers from any background are welcome.
• Platform engineering in large, regulated enterprise environments.

PLEASE DO NOT APPLY IF

• You are an engineering manager or director and your hands-on coding effectively stopped years ago. This role has no direct reports and is measured on what you build.
• You are a generic solution or cloud architect who produces designs and governance but does not implement, review code or debug production systems.
• You are a data scientist or ML researcher without strong production software engineering behind you.
• Your Python is mainly scripting, automation or notebooks rather than production services and libraries that other teams depend on.
• Your Generative AI experience is proof-of-concept, pilot or demo work that never carried real production traffic or had its quality measured.
• You have consumed SDKs and frameworks but never owned one through design, packaging, versioning, compatibility and deprecation.
• You cannot sustain roughly 6 hours of daily overlap with US Central Time, or cannot be based in Pune or Chennai after the initial 12 months.

SELECTION PROCESS

A GenAI engineering skill test, followed by technical and behavioural interviews. The screening questions on this application are mandatory: applications without them cannot be assessed.