Apply: Senior AI Engineer, RBC Studio, Kochi

Senior AI Engineer, Kochi

RBC Studio

I run a small software studio. We build apps and AI tools for clients in the US, and one of our biggest projects is a CRM for a healthcare company that helps people in addiction recovery.

I need one senior engineer in Kochi who can take charge of that project and a few others. You'd be the main technical person, so I need someone who can make decisions, run with them, and tell me straight when something won't work.

We move quickly. Priorities can change in the middle of a week, and we ship small updates often instead of waiting for big launches. If you've worked at a startup, you'll know what that's like.

I also expect you to use AI tools every day. Claude Code, Cursor, whatever you prefer. Good engineers today get through a week's work in a day or two with these tools, and that's the speed we're after.

What you'll build

The big part of this job is building our own AI, end to end. We work with patient data, so a lot of it can't go to outside AI companies. You'd fine-tune open-source models like Llama, Qwen or Mistral, host them securely on our own AWS servers, and scale them as usage grows.

On top of that you'd build a set of AI agents for the client's staff:

• One that reads calls, emails and transcripts and keeps the CRM up to date
• One that helps with insurance claims, denials and appeals, and pulls in the clinical team when needed
• One that supports admissions and intake
• One that checks in on alumni after treatment and connects them to a therapist when something is wrong

Every agent suggests and a person approves. You'd own all of it, from the first design to running it live.

What you'll do day to day

• Design and build the AI agents, and the screens staff use to review and approve their work
• Prepare training data from real calls, emails and documents, and strip out patient details first
• Fine-tune and test models, and prove each new version is better than the last
• Host models on private AWS servers, keep them secure, and scale them without the bill running away
• Keep working on the CRM itself (React frontend, Node.js and Express backend, PostgreSQL)
• Talk directly with me and the US client, and push back when something is a bad idea

Experience

• 7+ years in software engineering, with some of it leading projects without a manager checking your work
• 2+ years building with AI models in a live product, not just in a notebook
• You've fine-tuned open-source models yourself and can show me what you built
• You've hosted and scaled a model for real users, and you know what it cost to run
• You've worked with sensitive data (healthcare, finance or similar) and know what keeping it private takes in practice

Technical skills

Open-source models and training

• Llama, Qwen, Mistral or similar
• Fine-tuning with Hugging Face, LoRA and QLoRA, PEFT, Axolotl or Unsloth
• Preparing data for training: cleaning it, labelling it, and removing personal details
• Testing models: building test sets, scoring before and after, and catching made-up answers

Agents and search

• Building agents that work together, with a human approval step (LangGraph, LlamaIndex or similar)
• Search across documents (RAG) and vector databases such as pgvector, Qdrant or OpenSearch
• Speech-to-text with Whisper or similar, and working with messy call transcripts and emails

Hosting, security and scale

• Serving models with vLLM, TGI or Ollama
• AWS: EC2 GPU servers, EKS or ECS, SageMaker, S3, RDS
• Private networks with no public internet access, encryption, access controls, and a log of every action
• Docker and Kubernetes, scaling up and down with demand, and keeping GPU costs under control
• Monitoring model speed, errors and quality once live

Application

• Python
• React (Next.js is a plus)
• Node.js and Express, Prisma
• SQL and PostgreSQL
• Git, and deploying through GitHub Actions

Compliance

• HIPAA, and how it applies to AI and data storage

Bonus

• Healthcare experience, especially behavioral health
• 42 CFR Part 2 (US privacy rules for addiction records)
• Insurance billing, claims or revenue cycle work
• Anything public I can look at, like GitHub or Hugging Face

Hours and location

Based in our Kochi office. You can work from home up to 2 days a week. Good spoken English, and you're happy working 2pm to 11pm IST so we overlap with our US clients.

Pay

Competitive, plus a bonus. This role has a clear path to leading our tech team.

How to apply

Click Start below. It takes about 30 to 35 minutes: a few quick questions, your CV, two short written answers, and a 15 to 20 minute voice conversation with our AI interviewer. When you talk about your work, be ready to tell us about one model you fine-tuned or deployed: what it did, how you measured it, where it ran, and what it cost per month.