Prakyat Prakash
Open to 2026 SWE & ML roles
Hi, I am Prakyat Prakash

I build systems that work
and models that learn.

Software engineer and pipelines architect by practice: data pipelines, full-stack systems, and ML models that go from notebook to production. I also do research, with a focus where computation meets biology.

Prakyat Prakash
Education
RIT

Rochester Institute of Technology

Aug 2024 – Dec 2026
Master of Science, Computer Science · Rochester, NY
Relevant coursework
Data Structures & AlgorithmsComputer ArchitectureMachine LearningFoundations of AIBig DataAdvanced Computer Vision
Research

Where computation meets biology

My research sits at the intersection of machine learning and molecular biology, specifically how computational methods can extract signal from messy, large-scale biological data. I work in the Cui Lab at RIT, where the problems range from proteomics data engineering to understanding how the genome regulates itself.

Right now I'm focused on two things: building rigorous PTM extraction pipelines for post-translational modification data, fixing assumptions the field has accepted for years, and studying genomics and 3D chromatin organization through deep learning models. The thread connecting all of it is the same: biology generates enormous, noisy datasets, and good software and good models are what turn that noise into biology.

Published work

Machine learning-based determination of sex-related bladder cancer biomarkers

Pizzi, J.R., Adhikari, I., Prakash, P., Miyamoto, H., & Cui, F. · Frontiers in Bioinformatics, 2026

Intrinsic DNA codes govern distinct modes of nucleosome–transcription factor interactions

Carson, C.W., Nagalakshmi, S.U., Adhikari, I., Freewoman, J.M., Pizzi, J.R., Prakash, P., & Cui, F. · Nature Communications, 2026
Experience

Where I've been building

May 2025 – Present
Rochester, NY

Graduate Research Assistant · Machine Learning

Rochester Institute of Technology
  • Co-authored 2 research papers in computational genomics: a peer-reviewed publication in Frontiers in Bioinformatics on ML-based bladder cancer biomarker discovery, and a paper in Nature Communications on nucleosome–transcription factor interactions.
  • Engineered an end-to-end PTM verified-negatives extraction pipeline processing 24.4M raw peptide-spectrum matches into 361,789 verified phosphosites across 35 parallel workers.
  • Built a dual-pass FASTA verification pipeline with tryptic-context filtering, correcting trypsin-cleavage bias and reducing terminal-K over-representation from 33.2% to 3.6%.
Jan 2025 – Apr 2025
Remote, NY

AI Engineer

Handshake AI
  • Architected an LLM-based RAG pipeline for Python automation tasks, improving code generation accuracy by 25% through targeted retrieval optimization and prompt engineering.
  • Built a systematic LLM evaluation and debugging workflow using LangChain tooling, cutting output error rates by 30% and enabling faster root-cause diagnosis of model failures.
  • Standardized response generation frameworks for chat-based LLM interactions, driving consistent, production-grade execution across automated workflows.
Mar 2024 – Jun 2024
Bengaluru, India

AI Intern

Acinonyx Technologies Pvt. Ltd.
  • Built an AI-driven backend service with Flask and REST APIs to automate license management, cutting manual oversight by 30% and streamlining renewals for 500+ users.
  • Developed an ML model to predict license-expiration patterns, enabling real-time tracking and cutting unexpected service downtime by 40%.
  • Applied AI-powered analytics on transactional data to trigger automated payment reminders, improving processing efficiency and driving sales growth.
Projects

Things I've built

PTM Extraction dashboard, dataset summary

PTM Dataset Extraction Pipeline

Reconstructed phosphorylation and lysine-acetylation training datasets from 5 mass-spectrometry repositories, processing 24.4M peptide-spectrum matches into 361,789 tryptic-verified phosphosites. Replaced the field-standard "all non-positives are negatives" assumption with true experimentally-observed negatives via dual-pass FASTA verification across 35 parallel workers.

PandasNumPyUniProt
Dec 2025 – Mar 2026● Live
WC2026 Hydration Break Analyzer dashboard

WC2026 Hydration Break Effect Analyzer

Built a Python pipeline scraping every WC2026 match from ESPN's API, detecting exact hydration break timestamps and measuring attacking pressure in the 10 minutes either side. Found that 85–87% of pressing teams lose momentum after breaks while non-pressing teams consistently gain it — a statistically significant equalizer effect across 18+ matches. Dashboard auto-updates daily via GitHub Actions.

PythonESPN APINext.jsGroqGitHub Actions
Jun 2026● Live
CareBridge cross-border healthcare cost comparison dashboard

CareBridge

Cross-border healthcare cost navigator: sourced and structured pricing, treatment duration, and recovery data across 10+ countries for 30+ medical procedures. Built a RAG-based conversational chatbot answering patient queries on cost and hospital options, cutting average research time for cross-border care decisions by an estimated 40%. Modeled the full patient decision journey in a Neo4j knowledge graph spanning 5+ touchpoints, from initial inquiry to final hospital selection.

PythonRAGNeo4jLLMsVector DB
Ongoing2026
Get in touch
prakyat02@gmail.com
Open to software engineering and machine learning roles for 2026.
© 2026 Prakyat PrakashRochester, NY