AI/ML Engineer

Fitsol Ltd; Gurgaon; On-site; INR 2,500,000 - 4,500,000; Full time

Job summary

Fitsol in Gurgaon is hiring an AI/ML engineer to build the AI layer across our products — carbon accounting, ESG intelligence, and packaging design. You will work on real customer data and ship to production.

You will develop an AI assistant to extract data from supplier files for emissions calculations, implement RAG pipelines over ESG frameworks, and contribute to 3D packaging tools. The role requires hands-on with LLMs and data wrangling.

Qualifications

• 3+ years in production ML/AI.
• Strong Python and backend skills (FastAPI or similar).
• Hands-on with Vertex AI, OpenAI, Anthropic, or Bedrock — prompts, function calling, structured outputs.
• RAG end to end — chunking, embeddings, vector stores, retrieval evaluation.
• Computer vision basics — OpenCV, PyTorch or TensorFlow, object detection and segmentation.
• Web scraping at scale — Scrapy, Playwright, or Puppeteer.
• Data wrangling — entity resolution, unit normalisation, schema mapping.

Responsibilities

• Build an AI assistant that reads supplier files (invoices, utility bills, ERP exports, logistics manifests) and extracts data for Scope 1, 2, and 3 emissions calculations.
• Build RAG pipelines over GHG Protocol, emission factor libraries, and ESG frameworks (CDP, DJSI, EcoVadis, BRSR).
• Work on the packaging design tool — computer vision for nesting and overlap detection from 2D images, and 3D modelling for volume, material, and pack-out estimation.
• Build scrapers to fetch emission factors, regulatory data, and public ESG disclosures. Handle anti-bot, pagination, and schema drift.
• Build the data layer that makes scraped and extracted data usable — entity resolution, unit normalisation, schema mapping, confidence scoring, and human review queues.
• Set up evals and monitoring for every model and prompt we ship.

Skills

Python

Backend development

FastAPI

LLMs in production

RAG pipelines

Web scraping

Data wrangling

Computer vision basics

Vector stores

Tools

Vertex AI

OpenAI

Anthropic

Bedrock

Job description

Company: Fitsol Location: [Gurgaon / Hybrid / Remote] Type: Full-time Experience: 3-5 years

About Fitsol

Fitsol is a climate-tech startup helping enterprises measure and reduce supply-chain emissions. We work with manufacturers, automotive, and logistics companies across carbon accounting, ESG reporting, and packaging.

The role

We are hiring an AI/ML engineer to build the AI layer across our products — carbon accounting, ESG intelligence, and packaging design. You will work on real customer data and ship to production.

What you'll do

• Build an AI assistant that reads supplier files (invoices, utility bills, ERP exports, logistics manifests) and extracts data for Scope 1, 2, and 3 emissions calculations.
• Build RAG pipelines over GHG Protocol, emission factor libraries, and ESG frameworks (CDP, DJSI, EcoVadis, BRSR).
• Work on the packaging design tool — computer vision for nesting and overlap detection from 2D images, and 3D modelling for volume, material, and pack-out estimation.
• Build scrapers to fetch emission factors, regulatory data, and public ESG disclosures. Handle anti-bot, pagination, and schema drift.
• Build the data layer that makes scraped and extracted data usable — entity resolution, unit normalisation, schema mapping, confidence scoring, and human review queues.
• Set up evals and monitoring for every model and prompt we ship.

What we're looking for

Must-have

• 3+ years in production ML/AI. At least 1 year with LLMs in production.
• Strong Python and backend skills (FastAPI or similar).
• Hands-on with Vertex AI, OpenAI, Anthropic, or Bedrock — prompts, function calling, structured outputs.
• RAG end to end — chunking, embeddings, vector stores, retrieval evaluation.
• Computer vision basics — OpenCV, PyTorch or TensorFlow, object detection and segmentation.
• Web scraping at scale — Scrapy, Playwright, or Puppeteer.
• Data wrangling — entity resolution, unit normalisation, schema mapping.

Nice-to-have

• 3D geometry (Trimesh, Open3D, Three.js) or CAD formats.
• Agentic workflows (LangGraph, LlamaIndex or else).
• Interest in ESG, GHG Protocol, or carbon accounting. We will teach the domain.