BHANUHARSHA Y
MS COMPUTER SCIENCE · AI/ML ENGINEER · FULL STACK
ABOUT
I'm a graduate Computer Science student at Northeastern focused on machine learning, generative AI, and MLOps. I build production systems end to end — from LLM and RAG pipelines to full-stack web apps. My work spans published research on prompt optimization and hands-on engineering across data, models, and infrastructure. I care about systems that are fast, measurable, and reliable under load.
Northeastern University
MS, Computer Science
GPA 3.85 / 4.0 · Boston, MA
GITAM University
BTech, Computer Science & Engineering
GPA 3.80 / 4.0 · Visakhapatnam, India
- /AWS Certified AI Practitioner
- /Machine Learning with Python — IBM
- /Text Retrieval & Search Engines — UIUC
EXPERIENCE
AI Software Developer Intern
Aosenuma · Boston, MA
- /Architected a unified infra/LLM observability platform with OpenTelemetry, LiteLLM, Langfuse, Prometheus, and Grafana for end-to-end tracing and evaluation of production AI apps.
- /Engineered role-based authorization and data-access controls across Next.js, FastAPI, and PostgreSQL/Supabase for a secure multi-tenant case-management platform.
- /Built an automated LLM evaluation framework using RAGAS, golden datasets, and CI-integrated regression tests to catch quality regressions before deployment.
Research Assistant — Generative AI
InsightX Research Lab · Northeastern University
- /Built a fine-tuning-free prompt optimization system with DSPy and GPT-4o-mini to generate counter-narratives against online hate speech; ranked 3rd at LT-EDI@ACL 2026, with paper published at ACL Anthology.
- /Designed a 16-metric reward framework across 6 categories and integrated Detoxify, RoBERTa-MNLI, SBERT, BERTScore, and GPT-2 for automated, consistent counter-speech quality assessment.
- /Ran iterative Llama-3.1-8B prompt optimization experiments on Northeastern's Discovery HPC cluster (A100/V100 GPUs).
- /Benchmarked counter-narratives against Multitarget-CONAN using the NAACL 2024 LLM-CN-Eval framework and GPT-4o-as-a-judge across 8K+ hate-speech/CN pairs.
ML & Data Science Intern
Fluentgrid Limited · Visakhapatnam, India
- /Engineered an NLP chatbot with TensorFlow/Keras and NLTK, training a neural network across 35+ intent categories with 150+ patterns for partner detection and lead capture.
- /Automated data preparation with Tabula-py and pdfplumber to parse multi-section regulatory PDFs, cutting manual processing time by 70%.
SKILLS
PROJECTS
AutoMend
Incident response platform that detects ML infrastructure failures and runs approved remediation workflows.
Intelligent Traffic Management System
YOLOv7 traffic controller that changes lane timers from live vehicle density and emergency detection.
GUARD-RAG
Financial QA system that sends only risky questions into debate and improves TAT-QA F1 by 12.1%.
Kambaz LMS
Canvas-style LMS with role-based access, quizzes, enrollments, and 43 REST endpoints.
Store Sales Forecasting
Retail sales forecasting on 3M+ rows using lag features, rolling windows, and boosted trees.
Calendar Application
Java calendar app with time zones, recurrence, conflict detection, CSV import, and 100+ tests.
WANT THE DETAILS?
Download my résumé for the full breakdown of experience and projects.