$ cat about.txtHi, I'm Arjun Vankani — an AI Engineer working across Generative AI, Computer Vision, and NLP, currently exploring LLaMA fine-tuning, diffusion models, and CLIP-based retrieval.
- 🎓 M.Tech ICT (Machine Learning), Dhirubhai Ambani Institute of ICT —
2022–2024 - 🧪 Currently exploring: LLaMA fine-tuning, diffusion models, CLIP retrieval
- 📘 Book chapter provisionally accepted: Cybernetic Shield: Securing the Future of Machine Intelligence, 2025
- 🥇 Shri Dewang Mehta IT Award — Top Ranker, B.E. (2022)
- 💬 Fun fact: guest lecturer who still debugs shape mismatches at 2 AM
class Arjun:
def __init__(self):
self.role = "AI Engineer & Aspiring Research Scholar"
self.focus = ["Generative AI", "Computer Vision", "NLP"]
self.currently_exploring = ["LLaMA fine-tuning", "Diffusion models", "CLIP retrieval"]
def say_hi(self):
return "Let's build something intelligent 🚀"
|
Built a U-Net based dehazing pipeline, outperforming DCP, COA, FFA & AOD baselines. Synthetic haze generated via beta/airlight tuning for light-to-heavy conditions.
|
CNN + Deep SORT tracking, Omni-Scale feature extraction, and FaceNet for facial recognition — chained into a full re-identification pipeline.
|
|
Fine-tuned LLaMA into a domain-aware ChatGPT clone using LangChain embeddings for context-aware retrieval, evaluated with BLEU score.
|
Text → Image retrieval engine using CLIP embeddings, searchable via CLI or local folders. Query "rainy road at night" and watch it find it.
|
|
CSV-driven pipeline that auto-routes tasks (upscale / text2img / video / segmentation / StyleGAN) to ComfyUI — one click, zero manual wiring.
|
Gradio app bundling sentiment analysis, NER, Q&A, object detection/segmentation, and Stable Diffusion scene generation for environmental data.
|
Applied ML in Python – U. Michigan Python for Data Science – IBM GenAI with LLMs – AWS
ML Specialization – Stanford Neural Networks & Deep Learning – DeepLearning.AI Data Analysis with R – Google
