Skip to content
View AakashKotha's full-sized avatar
🎯
Focusing
🎯
Focusing

Highlights

  • Pro

Block or report AakashKotha

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don't include any personal information such as legal names or email addresses. Markdown supported. This note will be visible to only you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
AakashKotha/README.md

Hi there 👋

AakashKotha

I’m V V S Aakash Kotha, a Software Engineer and AI Systems Researcher pursuing my M.S. in Computer Science at the University of California, Davis.

I specialize in building LLM-driven and data-centric systems that merge scalable software design with applied AI research, from backend architecture to full-stack product engineering.


💻 Software Engineering Highlights

  • TinkerfAI – Full-Stack LLM Platform: Designed a production-ready zero-code ML training app using React + FastAPI + AWS (App Runner, Cognito, DynamoDB, S3); implemented guardrailed GPT-4o prompting and multi-role access workflows.
  • Retrieval-Augmented Generation – Wang Lab: Architected a cloud-based RAG system with topic-segmented vector stores, query decomposition, and Raptor summarization, reducing retrieval latency by 70%.
  • Backend Systems & DevOps: Built CI/CD pipelines with Docker and AWS App Runner; deployed multi-service APIs on AWS with container orchestration and secrets management.

🔬 Research & Applied AI

  • Silent Speech Decoding – Miller Lab: Synthesized an sEMG → speech pipeline (HuBERT + Tacotron-2) restoring laryngectomy patients’ voices with 82% intelligibility.
  • LLM Evaluation – Turing: Built a multi-agent benchmarking framework with LLM-as-a-Judge and consensus validation, auditing GPT-4o vs Gemini 2.5 Pro for reasoning reliability.
  • LLM Hallucination Audit: Designed a quantitative framework analyzing hallucination behavior using chi-square tests, z-tests, and logistic regression, revealing a 37% reduction in GPT-4o hallucination rate compared to GPT-3.5.

🧠 Technical Stack

Languages: Python, Java, C++, R, MATLAB, SQL, JavaScript, TypeScript
Frameworks & Libraries: React, FastAPI, Flask, Streamlit, PyTorch, TensorFlow, Keras, Scikit-learn, LlamaIndex, LangChain, LangGraph
Tools & Platforms: Docker, CI/CD, AWS (S3, DynamoDB), MongoDB, MySQL, PostgreSQL, Postman, Selenium, Git, Tableau
Concepts: System Design, OOP, Software Architecture, Deep Learning, NLP, LLMs, MCP, Computer Vision


🏆 Achievements

  • ECE Fellowship – $5,000 for contributions to agentic AI systems
  • AWS AI & ML Scholarship Recipient
  • UC Davis Teaching Assistant – taught 400 + students through live debugging and personalized guidance

🤝 Let’s Connect

📫 vkotha@ucdavis.edu
🔗 LinkedIn | GitHub

AakashKotha's streak

🧠 Interests

Software Engineering System Design AI Systems Machine Learning Deep Learning Natural Language Processing LLM Applications Data Science

Pinned Loading

  1. Unveiling-the-npm-Ecosystem-STA220-project Unveiling-the-npm-Ecosystem-STA220-project Public

    Bar plot showing Top 20 Countries by Contributor Percentage: https://aakashkotha.github.io/npm-contributors-map/ Bar plot showing Top 20 Countries by Contributor Percentage (among popular packages)…

    Jupyter Notebook

  2. SE_NPM_packages SE_NPM_packages Public

    Forked from travaditaher/SE_NPM_packages

    Jupyter Notebook

  3. Co.Heal Co.Heal Public

    CoHeal - Covid19 Help, Info & Tracker

    HTML

  4. Social_Distance_Detector Social_Distance_Detector Public

    This project will help you to detect social distancing in this COVID-19 situation at your premise. Using technology, more specifically the ML-based, I developed a mechanism, which can be implemente…

    Python 1

  5. VITHostelServices VITHostelServices Public

    Forked from VinayEdula/VITHostelServices

    JavaScript 1

  6. AakashKotha.Portfolio AakashKotha.Portfolio Public

    HTML