AI systems · full-stack products · applied ML
BITS Pilani, Goa
Mapping the invisible with code.
An engineering student focused on ML/AI engineering and full-stack product development. I build products where AI has a job to do — turning messy real-world data into clear, usable systems.
My work spans AI finance tools, neuromorphic image processing, biomedical signal prototypes, and LLM evaluation.
scroll to explore
A mix of medical ML, neuromorphic hardware, embedded avionics, and shipped product. Each one taught me something I'm still using.
A full-stack finance dashboard that turns PDF statements into spending patterns, editable categories, financial plans, and Gemini-powered insights.
Research project on energy-efficient image edge detection using spintronic devices and hardware-aware logic. Achieved +143% improvement over SAM2 baselines.
Low-cost digital stethoscope for ML-based cardiac abnormality classification using hybrid ensemble methods.
Real-time, calibration-free BP prediction from ECG/PPG signals using 1D CNN and Time Series UNet with lightweight PTT algorithm.
Embedded C++ firmware for I²C-connected baro/temp/accel sensors with Kalman-filtered altitude estimation for student rocketry project.
Benchmarked open-source LLaMA models against Gemini ground truth for educational Q&A using RAGAS + BLEU evaluation frameworks.
I'm Prajakta, an engineering student at BITS Pilani, Goa, interested in ML/AI engineering and product-minded software development.
Most of my work starts with messy data: PDF statements, model responses, biomedical signals, image datasets, or sensor readings. I enjoy turning that ambiguity into something structured — a dashboard, a model, an evaluation pipeline, or a product feature that helps someone make a better decision.
I'm especially interested in AI systems that are useful beyond the demo: tools with clear interfaces, grounded outputs, and a reason to exist.
“I like systems that make hidden patterns visible.”
Leading a 10-engineer avionics team within a 25-person student rocketry project at BITS Goa. Managing design and integration of flight computers, sensors, and telemetry systems.
AI Research Intern
Benchmarked open-source LLMs (LLaMA family) for educational Q&A using RAG pipelines. Evaluated with RAGAS + BLEU; integrated LlamaIndex for retrieval.
Product Lead
Drove product strategy and AI feature design with emphasis on user experience. Led competitive analysis and mapped strategic decisions based on insights.
Data Analysis Intern
Analyzed solar CME datasets to uncover correlations for space weather prediction. Time-series analysis on preprocessed NASA data.
Summer Intern
Designed 1D CNN + Time Series UNet models for blood pressure estimation. Engineered lightweight PTT-based algorithm for real-time BP prediction.