Chirag Goyal

Data Scientist

Bengaluru, Karnataka, India2 yrs 8 mos experience
AI EnabledAI ML Practitioner

Key Highlights

  • Expert in building scalable AI and ML solutions.
  • Proficient in advanced NLP applications and reinforcement learning.
  • Strong mathematics background enhances problem-solving capabilities.
Stackforce AI infers this person is a Data Science and AI professional with a focus on scalable solutions and advanced algorithms.

Contact

Skills

Core Skills

Reinforcement LearningRecommender SystemsNatural Language Processing (nlp)OptimizationGenerative AiMachine LearningArtificial Intelligence (ai)Data ScienceWeb ScrapingData Analytics

Other Skills

RankingGenerative AI ToolsDPOLLMRLHFLoRAFine TuningGraph DatabasesLSTMPyTorchObject-Oriented Programming (OOP)MATLABPython (Programming Language)TensorFlowTelemetry

About

I am an AI Engineer / Data Scientist with hands-on experience in building scalable AI and ML solutions, including LLMs, GenAI systems, advanced NLP applications and reinforcement learning. My work spans end-to-end model development, from data preprocessing to deployment, leveraging cutting-edge frameworks and cloud platforms. I bring strong problem-solving skills honed through proficiency in Data Structures and Algorithms (DSA) and a solid foundation in computer science fundamentals, including Object-Oriented Programming (OOP) and Operating Systems (OS). Additionally, I have a strong mathematics background in Probability, Statistics, Linear Algebra, and Calculus, which allows me to develop effective solutions to complex problems. Professional Highlights: • Fine-tuned LLMs using RLHF, LoRA, and DPO techniques, achieving significant performance improvements. • Developed Agentic RAG based Hybrid Recommendation systems. • Created scalable RAG-based applications with LangChain, Pinecone/Weaviate/FAISS, Knowledge Graphs and AWS Bedrock. •Worked extensively with transformer-based architectures and attention mechanisms. • Worked with Multimodal LLM usecases for conversational tasks. • Optimized Deep learning models with pruning and quantisation. • Developed and optimized traditional ML Regression models, such as Random Forests, XGBoost, and Logistic Regression. • Extensively worked on EDA and feature engineering. • Worked with clustering algos like K-means, DBSCAN & GMM. • Built Deep Learning systems using LSTMs, RNNs, and Transformer architectures. • Designed Deep Reinforcement Learning models for optimization use cases. • Developed end-to-end pipelines and dynamic model deployment using best MLOps practices. Skills: * Languages: Python, Java, SQL * Development: Machine Learning, Deep Learning, Reinforcement Learning, NLP, GenAI, LLMs * Frameworks & Tools: PyTorch, TensorFlow, Scikit-learn, LangChain, Langgraph, CrewAI, Docker, CI/CD, MLflow * Traditional ML: Regression, Classification, Ensemble Models, Hyperparameter Tuning * Deep Learning: LSTM, RNN, GRU, Attention Mechanisms, Transformer Architectures * NLP & Transformers: BERT, GPT, Llama, Hugging Face Libraries, Text Classification, Sentiment Analysis * Data Engineering: ETL Pipelines, MongoDB, MySQL, RAG Systems, Neo4j * Cloud Platforms: AWS,Azure * Other: Git, REST APIs, Selenium, Data Visualization (Matplotlib, Seaborn) ✉️ chirag.goyal.mail@gmail.com Last Updated: 7th Feb 2025

Experience

2 yrs 8 mos
Total Experience
1 yr 8 mos
Average Tenure
1 yr
Current Experience

Meesho

2 roles

Data Scientist - I

Jun 2025Present · 1 yr · Bengaluru, Karnataka, India · Hybrid

Data Science Intern

Apr 2025Jun 2025 · 2 mos · Bengaluru, Karnataka, India · Hybrid

  • Explored reinforcement learning based ranking system solutions for cold-start problem using LSTM actor-critic based temporal variant of DDPG, which is observed to improve item-level GMV ↑18% and IPV↑8.7% over CTR baseline by optimizing for long-term value.
Recommender SystemsRankingReinforcement Learning

Genpact

AI/ML Research & Development Intern

Jul 2024Dec 2024 · 5 mos · Gurugram, Haryana, India · On-site

  • Project 1:
  • Developed a Reinforcement Learning model to solve Vehicle Routing and Travelling Salesman Problem , reducing logistics costs by 15% , by utilizing LSTM encoder -decoder model with attention mechanism , integrated into actor -critic framework using Proximal Policy Optimization .
  • Utilized RL algorithms like Q- learning and SARSA on Sumolib networks for Route Optimization based on both time and distance .
  • Project 2:
  • Fine-tuned LLMs using RLHF according to client feedback data, achieving a 22% improvement over SFT (QLoRA).
  • Compared RLHF, SFT, and DPO fine-tuning on client feedback data, with DPO delivering a 17% performance improvement over Supervised fine-tuning (SFT) based on the customized metric.
Generative AI ToolsDPOReinforcement LearningLLMNatural Language Processing (NLP)RLHF+5

Mercedes-benz india

AI/ML R&D Intern

Jan 2024Mar 2024 · 2 mos · Bengaluru, Karnataka, India · Remote

  • Worked on computer vision techniques for improving the LiDar responses.
  • Did POC on building the prototype of AI agent in MB.OS (mercedes indigenous operating system)

Doit&c : department of information technology & communication, government of rajasthan(india)

Machine Learning Intern

May 2023Jul 2023 · 2 mos · Jaipur, Rajasthan, India · On-site

  • Project 1:
  • Automated RAG based news insights retrieval for Rajasthan govt by leveraging LangChain , OpenAI API, and FAISS DB for efficient search and retrieval .
  • Built a state news research tool enabling GPT-powered Q&A with source URLs enhancing governance information .
  • Project 2:
  • Improved the performance of the online news web scraping script by 12% using Selenium and BeautifulSoup .
  • Utilized Hugging Face transformer libraries to perform sentiment analysis and categorization on news articles .
Scikit-LearnMachine LearningBeautiful SoupTransformersPineconeNatural Language Processing (NLP)+6

Birla institute of technology and science, pilani

Research Collaborator

Feb 2023May 2023 · 3 mos · On-site

  • Designed and analyzed a protocol algorithm for optimally selecting a subset of L nanosatellites from a VLEO constellation.
  • Utilized Dijkstra's algorithm to optimize the selection criteria, significantly boosting link performance. Developed two models TBLP and TBLS to improve the efficiency of the satellite link selection process.
  • Designed an energy-efficient satellite network optimization model with TensorFlow, improving battery life by 129.52%.
  • Implemented routing algorithms with NetworkX, achieving 107.89% more residual energy & 0.51% less traffic blockage.

Team anant

Applied Scientist

Apr 2022Dec 2023 · 1 yr 8 mos

  • Team Anant is a group of passionate undergraduate students funded by ISRO, harbouring the dream of making indigenously built nanosatellite for multispectral remote sensing.
  • I led the AI vertical.
  • Project 1:
  • Led the implementation of the Telemetry and Communication Subsystem , ensuring reliable data flow for satellite communication .
  • Developed a Data Compression and Transmission System utilizing Huffman coding to optimize data for efficient transmission
  • Project 2:
  • Automated the FAQs regarding Anant test by using Google PaLM LLM model which reduced the manpower required for doubt -solving by 200 % .
  • Significantly reduced LLM Hallucinations with help of PromptTemplate , Utilized ChromaDB for querying and hosted the webapp on StreamLit.
PyTorchObject-Oriented Programming (OOP)MATLABPython (Programming Language)TensorFlowMachine Learning+1

Education

Birla Institute of Technology and Science, Pilani

B.E. — Electrical and Electronics Engineering

May 2020May 2025

National Academy, Alwar

Jan 2018Present

Hans International School

Jan 2020Present

Stackforce found 100+ more professionals with Reinforcement Learning & Recommender Systems

Explore similar profiles based on matching skills and experience