FlowCV Logo
Profile

AI intern interested in fine-tuning and benchmarking language models for clinical documentation. Eager to learn about evaluating models and building NER pipelines. Exploring error analysis, prompt iteration, and integrating solutions into live systems within different AI applications.

Internship Experience
  • •Engineered a Kotlin-based Android application for multicast audio streaming with <100ms latency using socket programming.
  • •Built a secure backend using Python and SQLite, hosted on AWS EC2 with basic CI/CD automation.
  • •Integrated Network Service Discovery for dynamic peer recognition, enabling auto-connect in under 3 seconds.
  • •Optimized packet transmission using adaptive link tuning, boosting throughput by ~70%.
  • •Validated system efficiency with NS3, Wireshark and other network protocol analyzer for debugging and root cause analysis.
    • Projects
  • •Developed a full-stack RAG-based application enabling users to chat with a diverse set of documents (PDFs, websites, CSVs, JSON, handwritten notes), processing up to 5+ source types concurrently.
  • •Engineered a robust backend pipeline for text extraction using PyPDF2, BeautifulSoup, and Google Cloud Vision API for OCR, incorporating session-based rate limiting at 25 pages/session.
  • •Integrated vector database (ChromaDB) and sentence-transformer embeddings(HuggingFace's BAAI/bge-small-en-v1.5) to power RAG and LangChain for LLM orchestration with Groq-hosted Llama3-70B.
  • •Built a real-time AI agent for interactive STT(Deepgram) and TTS(ElevenLabs), supporting web sessions (LiveKit) and phone calls (Twilio), with open-source LLMs.
  • •Utilized asynchronous programming and VAD(Silero) for low-latency, seamless user experience.
  • •Implemented metrics logging(TTFB, TTFT, EOU delay) to monitor and optimize performance.
  • •Built an LLM-driven platform for aptitude-based MCQ generation with contextual prompts using Flask backend and REST API endpoints.
  • •Enabled timed mode, real-time answer validation, and score tracking with latency <3 seconds.
  • •Deployed responsive UI using TailwindCSS and vanilla JavaScript with theme toggles and performance dashboard.
    • Skills
    Programming Languages:  Python, C++, C, Kotlin, Java
    Frameworks: FastAPI, Flask, Django, Android SDK, Jetpack Compose
    Relational & Non-relational Databases (SQL & NoSQL): MySQL, PostgreSQL, Redis, Firebase
    Tools: Git, GitHub, Android Studio, Postman, Docker
    AI/ML & LLMs: Supervised/Unsupervised Learning, Deep Learning, NLP, LangChain
    Coursework Subject:: Computer Networks | Database Management Systems | Object-Oriented Programming (OOPs) | Operating Systems | Data Structures & Algorithms
    Achievements
  • •Selected for Keploy API Fellowship(1,000 selected out of 18,500 applicants), gaining hands-on experience in open-source and API testing aligned with GSoC requirements.
    • Education
    Kalinga Institute of Industrial Technology (KIIT), B.Tech in Electronics and Computer Science | CGPA (6th Sem) - 8.57
    2022 – 2026 | Bhubaneswar, Odisha