Skip to content
View Aditi2k5's full-sized avatar

Highlights

  • Pro

Block or report Aditi2k5

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 is supported. This note will only be visible to you.
Report abuse

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

Report abuse
Aditi2k5/README.md

Hi πŸ‘‹, I'm Aditi Prabakaran

image

AI/ML Researcher β€’ 3D Vision β€’ LLM Systems β€’ Neural Rendering

🧠 About Me

  • πŸŽ“ B.Tech CSE Pre-final Year
  • πŸ”¬ Research Intern @ IIIT Hyderabad (Neural Rendering, 3D Vision)
  • πŸ§ͺ Focused on Computer Vision, LLMs, and Probabilistic ML
  • ⚑ Interested in bridging research β†’ real-world AI systems

πŸ”¬ Research Work

🧩 Neural Rendering & 3D Vision

  • Developed CLIP-guided 3D Gaussian Splatting framework
  • Semantic scoring + opacity regularization + pruning
  • Improved reconstruction quality and removed transient artifacts

πŸ“Š Bridge Risk Estimation (AAIMB 2026)

  • Probabilistic framework using NGBoost
  • Modeled latent structural variables (stress/strain)
  • Risk-based prioritization using uncertainty estimation

πŸ€– Hallucination Mitigation in LLMs (IEEE INDIACOM 2026)

  • Applied LoRA + PEFT fine-tuning
  • Improved faithfulness up to +49.8%
  • Reduced hallucination via calibration optimization

βš™οΈ Experience

πŸ”Ή Research Intern β€” IIIT Hyderabad

  • Multi-view 3D reconstruction in dynamic environments
  • Gaussian Splatting + Vision-Language supervision

πŸ”Ή Summer Research Fellow β€” IAS

  • Benchmarked NeRF variants (Instant-NGP, SSDNeRF, RobustNeRF)
  • Optimized for compute efficiency vs accuracy trade-offs

πŸ”Ή AI Intern β€” Saarc Masts Tech

  • Built RAG pipelines (LangChain + FastAPI + WebSockets)
  • Designed production-ready LLM workflows

πŸ”Ή AI Intern β€” Himitsu Lab

  • Multi-agent LLM reasoning visualization (AntV X6)
  • AI meeting assistant (real-time transcription + summarization)

πŸš€ Projects

🌫️ Air Pollution Forecasting (SIH Finalist)

  • LSTM-based temporal modeling (multi-station fusion)
  • RΒ²: 0.92 (O₃), 0.88 (NOβ‚‚)

πŸ›’ Grocery Price Prediction (Amazon ML Challenge)

  • Time-aware forecasting with XGBoost
  • SMAPE: 48.91

🧰 Tech Stack


πŸ“ˆ GitHub Analytics


πŸ† Achievements

  • πŸ₯‡ Smart India Hackathon 2025 β€” Top 5 National Finalist
  • 🌏 Google AI for Impact Hackathon β€” Top 98 (APAC)
  • πŸŽ“ IAS Summer Research Fellowship Recipient

🌐 Connect With Me


⚑ Research Interests

  • Neural Rendering (NeRF, Gaussian Splatting)
  • 3D Scene Understanding
  • LLM Alignment & Hallucination Mitigation
  • Probabilistic ML & Uncertainty Modeling
  • Real-time AI Systems

⚑ Open to research collaborations, internships, and AI-focused opportunities

Pinned Loading

  1. Future-of-Jobs Future-of-Jobs Public

    TypeScript 1

  2. NewsForge NewsForge Public

    TypeScript 1 1

  3. ConFeval-Faithfulness-Evaluation-for-LLMs ConFeval-Faithfulness-Evaluation-for-LLMs Public

    Jupyter Notebook

  4. Reinforcement-Learning Reinforcement-Learning Public

    I document what I learn and code in RL

    Python

  5. MedMap MedMap Public

    Forked from Rahuletto/medmap

    MedMap is a cross-platform mobile health companion built with Expo and React Native. It helps users track vitals, maintain medical records, visualize medication-to-condition relationships, and quer…

    TypeScript

  6. Sahayak Sahayak Public

    JavaScript