Open to opportunities

Riya Dadlani

@riyadadlani

Message

Computer scientist focused on ML/AI engineering, backend systems, and full‑stack development.

India

Message

What I'm looking for

I seek roles where I can build production ML/AI systems and scalable backends, collaborate cross‑functionally, and drive measurable product impact in healthcare or fintech.

I am a computer scientist and ML/AI engineer with hands‑on experience building scalable backend systems, production AI models, and full‑stack applications. I have improved system efficiencies and model accuracies across healthcare, finance, and real‑time retrieval products, driving measurable impact for thousands of users.

My work spans research internships and industry roles where I designed APIs, optimized LLM/SLM pipelines for medical diagnostics, and created end‑to‑end platforms (React, FastAPI, PostgreSQL, TensorFlow). I combine strong technical skills with leadership experience in student organisations and open‑source contributions to deliver reliable, production‑grade solutions.

Experience

Work history, roles, and key accomplishments

Research Intern

IIT Madras

Jul 2024 - Dec 2024 (5 months)

Improved LLMs/SLMs for medical applications, increasing diagnostic accuracy by 20% across datasets of over 5,000 patient records and built AI-driven models for automated diagnosis and report generation, enhancing efficiency by 35%.

Machine Learning Deep Learning LLM Python Data Analysis Model Evaluation Research

Backend Developer

Stripe

Mar 2023 - May 2023 (2 months)

Designed and implemented scalable backend systems using Python and Java, improving processing efficiency by 30% and supporting over 50,000 monthly customer transactions; developed and optimized APIs to enhance system integration by 25%.

Python Java API Design Backend Development PostgreSQL Docker Scalability Performance Optimization

Prompt / AI Engineer

Arrowhead

Designed and optimized multi-turn conversational AI prompts for finance and customer-service bots, improving context retention and response accuracy by 35% and reducing hallucinations by 25% through dynamic context injection and persona conditioning.

Prompt Engineering Python Redis Faiss