AI Engineer specializing in AI, LLMs, and Quantum-AI systems, with experience in building energy-efficient Transformer architectures, neuromorphic models. Currently contributing to a DST-sponsored project at NIT Jaipur o…
AI Engineer specializing in AI, LLMs, and Quantum-AI systems, with experience in building energy-efficient Transformer architectures, neuromorphic models. Currently contributing to a DST-sponsored project at NIT Jaipur on AI-assisted robotics and battery management systems, driving innovation in sustainable and high-performance computing.
<p>Applied machine learning techniques and statistical methods to analyze sensor-based datasets (e.g., battery performance data), identifying patterns in charging/discharging cycles and improving predictive modeling of system behavior.<br>
Developed ETL pipelines for DOBOT CR3 and MG400 test benches, producing clean datasets for analytics.<br>
Coordinated cross-functional integration of battery hardware with automation scripts, enabling synchronized experiments and consistent data acquisition.</p>
<ul><li>Developed ETL pipelines for DOBOT CR3 and MG400 test benches, producing clean datasets for analytics. </li><li>Coordinated cross-functional integration of battery hardware with automation scripts, enabling synchronized experiments and consistent data acquisition. </li><li> Video Demonstration – Robotics Automation Setup - https://youtu.be/4Aygwmuk-TU. </li></ul>
<p>Contributed to RLHF, SFT and SRT projects (Hopper, Cypher Evals, Matcha Stem, Pref Ranking, Extensions V2 Log), enhancing generative AI training through code evaluation and task curation.<br>
Evaluated AI model outputs and engineered comprehensive feedback reports to advance large language model accuracy and reliability across diverse task sets.</p>
Contributed to RLHF, SFT and SRT projects enhancing generative AI training...
<ul><li> Developed an autonomous financial analysis application, Q-OptiFolio, to perform end-to-end cryptocurrency portfolio optimization using specialized AI agents. Implemented a hybrid classical–quantum workflow using QAOA (IBM Qiskit) to solve the combinatorial asset-selection problem, achieving a Sharpe ratio of 0.33, 4.9 % expected annual return, and 8.7 % volatility on an ETH-ADA-DOGE portfolio. </li><li>Tech Stack: CrewAI, Qiskit, Litestar, React/Next.js , Groq (LLM Inference), OpenAI GPT-OSS-120B, CoinGecko API, Tailwind CSS. </li></ul>
<ul><li>Architected a state-of-the-art GraphRAG pipeline that integrates semantic search with knowledge graph traversal (using NetworkX), significantly improving reasoning capabilities across complex, unstructured 10-K financial reports. </li><li> Engineered a hybrid retrieval system combining Qdrant vector search with LLM-based entity–relationship extraction, delivering citation-backed answers and achieving 80% RAGAS factual accuracy. Added a Natural Language-to-Query module enabling automated, accurate Plotly data visualizations based on user intent. </li><li>Built an interactive Streamlit chat app with real-time token streaming and optimized 4-bit quantization (BitsAndBytes) to efficiently run Mistral-7B. </li><li>Technologies: LangChain, Prompt Engineering, LlamaIndex, Hugging Face, Mistral-7B , Qdrant, Transformers, Streamlit, PyTorch. </li></ul>
نظام التوظيف التشغيلي المعتمد على الذكاء الاصطناعي أولاً، الذي يوحّد ويؤتمت دورة التوظيف بالكامل للمؤسسات الحديثة حول العالم.
د.جوب منطقة حرة ذ.م.م. 2026 © جميع الحقوق محفوظة
