Looks like you have JavaScript disabled. For the full Twine experience, you will need to re-enable it.

I am a Master of Science in Computer Science candidate at the University of Massachusetts Amherst with a strong background in machine learning and data engineering. My experience spans developing sophisticated AI frameworks such as Multi-Agent Debate systems and Retrieval-Augmented Generation pipelines, as well as building and optimizing large-scale data pipelines in financial services. I thrive on innovating and improving systems by combining advanced algorithms and practical engineering. Passionate about AI, natural language processing, and data-driven solutions, I have worked on projects that mitigate bias in LLMs, enhance code summarization, and facilitate financial knowledge retrieval. My work includes practical applications in both academia and industry, emphasizing measurable improvements and reliability in data processing and machine learning pipelines.…I am a Master of Science in Computer Science candidate at the University of Massachusetts Amherst with a strong background in machine learning and data engineering. My experience spans developing sophisticated AI frameworks such as Multi-Agent Debate systems and Retrieval-Augmented Generation pipelines, as well as building and optimizing large-scale data pipelines in financial services. I thrive on innovating and improving systems by combining advanced algorithms and practical engineering. Passionate about AI, natural language processing, and data-driven solutions, I have worked on projects that mitigate bias in LLMs, enhance code summarization, and facilitate financial knowledge retrieval. My work includes practical applications in both academia and industry, emphasizing measurable improvements and reliability in data processing and machine learning pipelines.

Anvitha Kannapu





I am a Master of Science in Computer Science candidate at the University of Massachusetts Amherst with a strong background in machine learning and data engineering. My experience spans developing sophisticated AI frameworks such as Multi-Agent Debate systems and Retrieval-Augmented Generation pipelines, as well as building and optimizing large-scale data pipelines in financial services. I thrive on innovating and improving systems by combining advanced algorithms and practical engineering. Passionate about AI, natural language processing, and data-driven solutions, I have worked on projects that mitigate bias in LLMs, enhance code summarization, and facilitate financial knowledge retrieval. My work includes practical applications in both academia and industry, emphasizing measurable improvements and reliability in data processing and machine learning pipelines.…I am a Master of Science in Computer Science candidate at the University of Massachusetts Amherst with a strong background in machine learning and data engineering. My experience spans developing sophisticated AI frameworks such as Multi-Agent Debate systems and Retrieval-Augmented Generation pipelines, as well as building and optimizing large-scale data pipelines in financial services. I thrive on innovating and improving systems by combining advanced algorithms and practical engineering. Passionate about AI, natural language processing, and data-driven solutions, I have worked on projects that mitigate bias in LLMs, enhance code summarization, and facilitate financial knowledge retrieval. My work includes practical applications in both academia and industry, emphasizing measurable improvements and reliability in data processing and machine learning pipelines.

Available to hire

I am a Master of Science in Computer Science candidate at the University of Massachusetts Amherst with a strong background in machine learning and data engineering. My experience spans developing sophisticated AI frameworks such as Multi-Agent Debate systems and Retrieval-Augmented Generation pipelines, as well as building and optimizing large-scale data pipelines in financial services. I thrive on innovating and improving systems by combining advanced algorithms and practical engineering.

Passionate about AI, natural language processing, and data-driven solutions, I have worked on projects that mitigate bias in LLMs, enhance code summarization, and facilitate financial knowledge retrieval. My work includes practical applications in both academia and industry, emphasizing measurable improvements and reliability in data processing and machine learning pipelines.

Skills

Experience Level

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Intermediate

Intermediate

Intermediate

Intermediate

Intermediate

Work Experience

Machine Learning Researcher at BioNLP Lab (CICS)

May 1, 2025 - August 22, 2025

Developed a multi-agent debate framework simulating expert annotator dynamics for complex Social Determinants of Health (SDOH) prediction tasks using GPT-4o and LLaMA models. Improved decision accuracy by integrating critique agents and consensus mechanisms. Reduced annotation costs through adaptive test-time compute and expert-student agent routing, including prompt optimization.

Machine Learning Researcher at Laser Lab (UMass)

September 1, 2024 - August 22, 2025

Architected a context-aware Retrieval Augmented Generation (RAG) system that enhanced automatic code summarization by incorporating callee method context via abstract syntax trees (AST) traversal, achieving 14% BLEU and 3.7% METEOR improvements. Improved prompt engineering by integrating callee signatures and comments, increasing comment generation metrics by 5.4% across multiple LLMs, surpassing no-context baselines.

Software Engineer, Information Services Group – C10 at Citi

October 1, 2022 - August 22, 2025

Built ETL data pipelines for multi-format data ingestion, batch standardization, and enrichment of large-scale equities data using modular Spark jobs, achieving 99% SLA compliance for downstream analytics. Developed a framework to process over 1 million daily equity transactions in FIX protocol, reducing latency by 25%. Engineered real-time streaming pipelines with Kafka Streams to handle triple transaction volume with optimized throughput and fault tolerance.

Software Engineer, Global Information Warehousing – C09 at Citi

September 1, 2021 - August 22, 2025

Developed interest statement generation workflows, reducing processing time by 30%. Deployed data lake architecture, cutting query latency by 15%. Evaluated and tested Unravel AI for automatic job failure detection and notifications across 100+ Spark, Hive, and Impala workflows, reducing manual intervention by 80%. Unified data archival into HBase for improved reprocessing and system reliability; validated over 10 million daily transactions with comprehensive quality checks.