I specialize in full stack software development, and I’m passionate about building Responsible AI and GenAI applications, leveraging deep expertise in Python, C++, Java, TypeScript, JavaScript, SQL, and QML, alongside robust cloud proficiency in AWS, GCP, Azure, and Databricks. I excel in GenAI technologies including Hugging Face Transformers, LangChain, GPT-3/4, Llama 2-3, Retrieval-Augmented Generation (RAG), LLMOps, FAISS, Pinecone, and LangSmith, as well as NLP frameworks such as spaCy, NLTK, AllenNLP, and FastAPI, complemented by speech recognition tools like Kaldi and DeepSpeech. My experience extends to machine learning and deep learning where i leverage TensorFlow, TensorFlow Lite, PyTorch, Keras, Scikit-learn, MLlib, MLOps, MLFlow and a variety of supervised and unsupervised algorithms. My expertise covers computer vision (OpenCV), recommendation systems (Surprise, LightFM, RecBole), and development whithin distributed systems environment leveraging PySpark, Hadoop, Apache Kafka. I am adept in data management and databases such as PostgreSQL, MongoDB, Neo4j, DynamoDB, Snowflake, and Elasticsearch, and proficient in data visualization tools like D3.js, Tableau, Power BI, and Plotly. My experience in DevOps methodologies with Git, Docker, Kubernetes, Terraform, Jenkins, and GitHub Actions ensures seamless CI/CD pipelines and scalable deployments. I have successfully led cross-functional teams in designing and deploying scalable AI solutions, including multi-agent systems, autonomous drones, predictive maintenance models, personalized recommendation engines, and real-time NLP applications. Leadership and awards: I was recognized for leveraging advanced technologies to enhance operational efficiency, accuracy, and user engagement, I have delivered transformative AI solutions that significantly boost customer satisfaction and ROI for industry leaders such as ExxonMobil, Nissan, Polaris, Airwise, CNHi, Versaware, and Sydaptic.

Designed and deployed LLM, NER, Predictive Modeling, ML, NLP, Personalized Recommendation, Computer Vision systems to fuel a proprietary AI-driven market opportunity analysis system for Rettera, processing over 2.5 TB of geospatial and multimedia data from 300+ sources. Fine-tuned an open-source LLM (BLOOM) using Hugging Face Transformers and DeepSpeed, achieving a 31% increase in market trend prediction accuracy and reducing model training time by 45%. Engineered a recommendation engine utilizing FAISS for high-dimensional vector search, Matrix Factorization techniques, and Collaborative Filtering, delivering personalized market insights with a 60% reduction in query latency. Developed NLP pipelines using LangChain, spaCy, and PySpark, automating the extraction of over 500k key financial entities with Conditional Random Fields (CRF), Named Entity Recognition (NER) techniques, and Support Vector Machines (SVM), resulting in a 58% improvement in data processing efficiency. Leveraged PySpark to optimize distributed data processing, reducing ETL time by 62% for datasets exceeding 1 billion records. Visualized insights through interactive Plotly dashboards, enhancing decision-making speed by 40% for investment strategies.Leadership & Awards: Led a team of 5, pioneering initiatives that increased market prediction accuracy by 31%. Received 7 prestigious awards and selected to join competitive accelerators, namely YC, MIT Fuse, MIT DeltaV, IDEAS, Sandbox, Start MIT, VMS, and MIT $100k.

Developed a wide range of NLP, Deep Learning, Bayesian Models, and ML models to enable modular applications for a Clinician Assistant Medical OS. I leveraged advanced AI and machine learning techniques to automate ICU processes, resulting in a 74% reduction in nurse documentation workload and a 94% decrease in medical errors during patient handoffs. Engineered an end-to-end NLU pipeline utilizing spaCy, NLTK for text preprocessing and entity recognition, Gensim for topic modeling, and FastText for word embeddings. Implemented predictive modeling with PyTorch and TF employing LSTM, GRU, and Attention mechanisms for time-series analysis of vital signs, achieving 90% accuracy in predicting patient deterioration. Integrated sensor fusion architecture to unify data from over 20 medical devices, utilizing a custom CNN-RNN hybrid model, DNN-HMM, and ensemble methods for real-time decision support, leading to enhanced patient safety. Deployed the solution on AWS used Docker and Flask, while managing data persistence with PostgreSQL and implementing CI/CD practices with Git and Jenkins. Leadership & Awards: Orchestrated a high-impact team, achieving a 74% reduction in documentation workload through advanced AI integration. Won Mass innovation startup award.