AI Engineer

Ekkomd; Chennai, Tiruchirappalli, Coimbatore District, Bengaluru, Hyderabad; On-site; INR 1,800,000 - 2,600,000; Full time

Job summary

Ekko Medical Solutions Private Limited is hiring an AI Engineer to own the LLM pipeline at the core of Ekko's healthcare product. You will design prompts, evaluate models, and build evaluation frameworks to ensure clinical notes and structured data are accurate.

You will report to the CTO and collaborate with backend, QA, and product teams in a fast-paced startup setting where ambiguity is common and shipping matters more than perfection.

Qualifications

• 3+ years of ML/AI experience in production with hands-on LLM work.
• Strong Python proficiency in production codebases.
• Proven prompt engineering experience focused on model reliability.
• A systematic approach to evaluating AI output quality.
• An accuracy-focused, pragmatic mindset backed by rigorous measurement.

Responsibilities

• Build and enhance the LLM pipeline stages that convert audio into clinical notes and structured data.
• Design, test, and version prompts with accuracy and safety front of mind.
• Evaluate and fine-tune models (e.g., OpenAI) for quality and cost.
• Build evaluation frameworks that compare AI outputs against ground-truth references.
• Partner with backend, QA, and product teams; report to CTO.

Skills

Production ML/AI

Python proficiency

Prompt engineering

Model evaluation

Startup mindset

Tools

RAGAS

LLM-as-judge

TypeScript/NestJS

Job description

Ekko is an AI-powered healthcare product that turns clinical conversations into structured documentation. Our pipeline transcribes audio, generates clinical notes, and extracts structured data using LLMs. We ship across three surfaces — a web app, a backend service, and a browser extension — used by clinicians who depend on accuracy. We're a small, early-stage team.

We're hiring an AI Engineer to own the LLM pipeline at the heart of Ekko's product — the stages that turn raw clinical audio into accurate notes and structured data. You'll design and version prompts, evaluate and fine-tune models, and build the evaluation frameworks that tell us whether the AI's output is actually correct, not just plausible. You'll report to the CTO and work closely with backend, QA, and product.

Pipeline Development

• Build and enhance the pipeline stages that convert audio into clinical notes and structured data
• Improve transcription accuracy across speech-to-text providers (AWS Transcribe, Google Speech, Azure Speech)

Prompt Engineering & Model Selection

• Design, test, and version prompts, with accuracy and safety front of mind
• Evaluate and fine-tune models from providers like OpenAI for quality and cost

Quality & Evaluation

• Build evaluation frameworks that compare AI outputs against ground-truth references
• Define and track clinically-relevant quality metrics

Cross-Functional Collaboration

• Partner closely with backend, QA, and product teams
• Report directly to the CTO

What You Bring

• 3+ years of ML/AI experience in production, with hands-on LLM work
• Strong Python proficiency in production codebases
• Proven prompt engineering experience focused on model reliability
• A systematic approach to evaluating AI output quality
• An accuracy-focused, pragmatic mindset backed by rigorous measurement
• Comfort with startup life: ambiguity, ownership, and shipping over perfection

Nice to Have

• Healthcare or clinical AI experience, or other high-stakes, accuracy-critical domains
• Speech-to-text / ASR systems expertise
• Familiarity with LLM evaluation tooling (RAGAS, LLM-as-judge)
• TypeScript/NestJS backend knowledge
• Compliance framework familiarity (HIPAA, NABH, DPDP)

Why join Ekko

• Real ownership. You're building the LLM pipeline that is the core of Ekko's product, not tuning a feature at the edges.
• AI product depth. You'll work on production LLM pipelines in a clinical setting where accuracy actually matters.
• Direct impact. Small team, fast feedback, decisions made in days not quarters.
• Good people. We take the work seriously and ourselves a lot less so.