About Innoterra :

Innoterra India Pvt Ltd operates high-velocity agri-tech and food businesses - fresh produce, dairy, animal nutrition, and distribution - managing procurement, quality, logistics, and last-mile supply chain at scale across India.

We are building an AI Decision Intelligence Platform that turns operational data into governed, real-time business decisions.

This is early-stage, high-ownership work. The platform and its direction are being shaped now, and the engineers joining at this stage will have a direct say in how it evolves - choosing tools, shaping architecture, and making decisions that will be in production for years.

Who We Are Looking For :

- Thinks analytically - breaks a problem down before writing code; asks why before asking how.

- Gets things done - ships working solutions without waiting for perfect requirements or perfect infrastructure.

- Owns the outcome - monitors what they build, fixes it when it breaks, and cares whether it works in production.

- Learns fast - picks up a new tool, library, or concept and applies it without hand-holding.

- Communicates clearly - explains technical decisions to non-engineers, and business tradeoffs to engineers.

- Has opinions - voices a reasoned point of view on tooling and approach - and is open to being wrong.

If you read documentation before asking questions, write a test before calling something done, and get genuinely curious when a pipeline produces the wrong number - this role is for you.

What You Will Do :

AI and ML Implementation :

- Work across the AI/ML lifecycle - data preparation, feature engineering, model training, evaluation, and production deployment.

- Train, fine-tune, and experiment with models across problem types : classification, regression, time-series forecasting, and optimisation.

- Work with large language models - prompt engineering, fine-tuning, and building grounded RAG workflows.

- Take models from experimentation into production : versioning, serving, monitoring, and iterating on real-world feedback.

- Contribute to decisions on model selection, tooling, and AI approach - and stay current enough to have informed opinions.

Data Pipelines :

- Design and build pipelines that move, transform, and validate data reliably from source systems into AI-ready formats.

- Work with batch and event-driven patterns - choosing the right approach for each use case.

- Ingest from internal operational systems and external feeds : market prices, weather APIs, commodity data.

- Implement data quality checks - schema, freshness, completeness, business rules - and handle failures gracefully.

- Write pipelines that are robust, observable, idempotent, and easy for the next engineer to pick up.

Deployment to Server :

- Package and deploy models, pipelines, and APIs to Linux servers and cloud-managed environments.

- Set up and manage model serving - REST endpoints, batch scoring jobs, and scheduled execution on remote servers.

- Containerise models and services using Docker for reproducible, portable deployments.

- Manage configuration, environment variables, and secrets cleanly across dev and production.

- Monitor deployed services - uptime, resource usage, logs - and respond to failures before the business notices.

- Automate repetitive deployment steps : start scripts, health checks, and restart-on-failure setups.

APIs, Outputs and Engineering Quality :

- Build Python REST APIs that expose model outputs as actionable decisions - scores, recommendations, confidence, explanations.

- Connect AI outputs to dashboards, operational apps, and messaging channels where business users already work.

- Build feedback loops that capture whether recommendations were acted on, overridden, or ignored.

- Instrument everything - structured logs, metrics, and monitoring for pipelines and models in production.

- Write tests, review code, detect drift early, and hold a high bar for what goes into production.

Technology :

We are an AWS-first team. Beyond that, we are pragmatic - we choose tools that fit the problem, not tools that sound impressive. You will be expected to have opinions on this.

Core - Be Comfortable Here :

- Python - the primary language for everything : model code, pipelines, APIs, and deployment scripts.

- ML and AI libraries - scikit-learn, pandas, NumPy, and at least one deep learning or LLM framework.

- Data pipeline orchestration - experience with any modern workflow tool; the specific choice is open.

- Server deployment - packaging, configuring, and running services on remote Linux environments.

- Docker - containerisation for reproducible, portable deployments.

- REST APIs - building and consuming them; FastAPI or equivalent Python framework.

- Linux - SSH, bash, process management, log reading; you work on servers, not just laptops.

- Git - branching, pull requests, and meaningful code review.

Relevant - Exposure or Willingness to Learn :

- Cloud storage and compute - AWS S3, EC2, Lambda, or equivalent managed services.

- Cloud AI/ML platforms - AWS SageMaker, Amazon Bedrock, or equivalent.

- LLM APIs - OpenAI, Bedrock, or similar; prompt engineering and structured output handling.

- Time-series forecasting libraries - Prophet, statsmodels, or equivalent.

- Vector databases - for RAG and semantic retrieval workflows.

- Optimisation libraries - for constraint-based decision and scheduling problems.

- CI/CD basics - automated testing pipelines and deployment automation.

What You Bring :

AI and ML - First Priority :

- Solid understanding of the AI/ML ecosystem : model training, evaluation, deployment, and monitoring in production.

- Hands-on experience across at least one problem type - classification, forecasting, NLP, or LLM-based systems - end to end.

- Familiarity with LLMs : what they are good at, where they fail, and how to use them responsibly in a business context.

- Ability to reason about which AI approach fits a problem and articulate why alternatives were not chosen.

Python and Data Engineering :

- 2-4 years of production Python - clean, tested, well-structured code you would defend comfortably in a review.

- Experience building data pipelines that run reliably in production, not just locally.

- Comfortable handling messy, incomplete, or late data - not just clean tutorial datasets.

Deployment and Operations :

- Comfortable deploying and running services on Linux servers - independently.

- Experience packaging and shipping code to a server : environment setup, configuration, and service management.

- Basic cloud familiarity - enough to navigate AWS, read logs, and understand where things run.

Good to Have :

- MLOps practices : model versioning, drift monitoring, or automated retraining pipelines.

- CI/CD exposure - automated testing and deployment.

- Domain exposure in agri-tech, FMCG, dairy, or food supply chain.

What We Offer :

- A direct say in technology and AI direction - not a back-row execution role.

- Real production problems across multiple business units at meaningful scale.

- A small, high-ownership team where contributions are visible and opinions are heard.

- Full-stack AI exposure : feature engineering, ML, LLMs, optimisation, deployment, and feedback loops.

- Competitive compensation for the right candidate.