I am a researcher and developer specializing in Deep Learning, with a deep focus on Natural Language Processing (NLP). I am currently pursuing a B.Tech in Computer Science and an M.S. by Research in Computational Linguistics at the International Institute of Information Technology (IIIT), Hyderabad.

My M.S. thesis research centers on advancing Machine Translation for low-resource Indian languages and exploring advanced alignment techniques like RLHF, including methods such as DPO and KTO. I am a Kaggle Grandmaster and an active NLP Blogger on Medium, where my articles have amassed over 300,000+ views.

• 🔬 Publishing research in top-tier NLP conferences like NAACL, WMT, EAMT, and MT Summit.
• 👨‍💻 Building practical, open-source AI/ML tools. Check out my codetoprompt library on PyPI!
• ✍️ Sharing my knowledge on advanced NLP topics on my Medium blog.
• 🏆 Passionate about competitive machine learning and problem-solving.

I am currently looking for impactful full-time roles in AI Research, Applied Science, or Deep Learning Engineering. I am driven to contribute to ambitious projects at the forefront of the AI field, particularly in areas like foundational models, generative AI, and agentic systems. If you have opportunities that align with my expertise, I would love to connect.

• arXiv 2025 (cs.CL): Crosslingual Optimized Metric for Translation Assessment of Indian Languages (COMTAIL). [paper]
• codetoprompt: Open-source CLI for converting codebases and web content into optimized LLM prompts - 20k+ PyPI downloads. [repo]
• Amazon (May–Oct 2025): Applied Scientist Intern : Made production-scale product embedding models.
• MT Summit 2025: Progressive Perturbation with KTO for Enhanced Machine Translation of Indian Languages. [paper]
• WMT 2024: Yes-MT’s Submission to the Low-Resource Indic Language Translation Shared Task. [paper]
• ICON 2024: Decoding Fake Narratives in Spreading Hateful Stories: A Dual-Head RoBERTa Model with Multi-Task Learning. [paper]
• SemEval 2024: Fine-tuning Language Models for AI vs Human Generated Text Detection. [paper]
• EAMT 2024: Assessing Translation Capabilities of LLMs involving English and Indian Languages. [paper]

Here are some of my latest projects that highlight my work in Deep Learning and NLP.

• codetoprompt
  A powerful CLI tool that transforms local codebases, GitHub repositories, and web content into a single, optimized prompt for LLMs. Features an interactive file selector, smart code compression, and multiple output formats.
  Check it out on PyPI.

• Test-Time Compute for SLMs
  An in-depth research project exploring test-time compute allocation (like budget forcing) to enhance reasoning in Small Language Models (SLMs). We benchmarked models like Phi-3 and Qwen-1.5B on advanced reasoning tasks (MATH, GPQA) using techniques like CoT/CoD prompting and GRPO fine-tuning.

• Multi-Agent Open-Domain Q&A System
  A sophisticated Q&A system built with specialized agents for query modification, document retrieval (sparse and dense), ranking, and answer generation to tackle complex, open-domain questions.

• RAG with Llama3 Agents
  An implementation of a Retrieval-Augmented Generation (RAG) pipeline leveraging Llama-3 and the ReAct (Reasoning and Acting) agent framework to provide accurate, context-aware responses from a knowledge base.

• Programming: Python, C/C++, Git, GitHub
• AI/ML Frameworks: PyTorch, TensorFlow, Scikit-Learn, Pandas, NumPy, Hugging Face
• Cloud & Big Data: AWS (SageMaker, EC2, S3), PySpark
• Databases: MongoDB, MySQL, PostgreSQL
• NLP/AI Domains: Large Language Models (LLMs), RAG, Agentic AI, Machine Translation, Fine-Tuning (SFT, RLHF [PPO, DPO, KTO], GRPO), Quantization, Information Retrieval, Vector Databases (Qdrant)

I'm always open to discussing new projects, research ideas, or collaboration opportunities. Feel free to connect with me!

• LinkedIn: linkedin.com/in/yash-bhaskar
• Medium: medium.com/@yash9439
• Email: [email protected]

What I cannot create, I do not understand. – Richard Feynman