Ctrl+K
-
1.52 kB
initial commit
-
209 Bytes
SHIVAM KUMAR [email protected] | 9871029576 | New Delhi, India linkedin.com/in/shivam-jha-3515a713a | github.com/kumar045 | twitter.com/@dk238347 SUMMARY To secure a position that leverages my expertise in AI-related skills, including machine learning, deep learning, time series forecasting, Bayesian statistics, Hugging Face transformers, LLM, diffusion, Langchain, and prompt engineering, along with my proficiency in optimization, agile development methodologies, GCP, PySpark, big data analysis, MERN stack, Docker, Kubernetes, IBM Cloud, and full stack cloud web development, to contribute to the success of an organization. EXPERIENCE (4 YEARS 2 MONTHS) Senior AI Software Developer 17 July 2023 - Present Sveltetech Gurgaon, Haryana • Contributing to Big Data Project, a Big Data and Generative AI project focused on processing, unstructured data analysis, generation, and RAG. • Fine-tuning Large Language Models with Multimodal Data (Text, Audio, Image Text) Enhanced LLMs by finetuning with multimodal data (text, audio transcriptions, OCR-extracted image text) to transform unstructured data into structured formats, optimizing storage and enabling advanced analytical workflows. • Experience working with Natural Language Processing, Django, PySpark, Apache Flink, Apache Kafka, Docker, LLM, Generative AI, Hugging Face Transformers, Elastic Search, Neo4j, and Langchain. Senior Artificial Intelligence Technical Specialist 24 Jan 2022 - 16 July 2023 Skill-Up Technologies Noida, UP • Developed content, labs, and assessments related to AI and Cloud technologies for IBM learning programs on Coursera, edX, and IBM’s in-house learning platforms as part of SME services. • Experience working with Computer Vision, Natural Language Processing, Machine Learning, Deep Learning, Django, MERN, Docker, LLM, Diffusion, Hugging Face Transformers, Terraform, Kubernetes and Langchain. Artificial Intelligence Technical Specialist 21 Jun 2021 - 24 Jan 2022 Skill-Up Technologies Noida, UP • Working as an AI and Cloud Subject Matter Expert for IBM learning programs on Coursera, edX, and IBM’s in-house learning platforms as part of SME services. Artificial Intelligence Engineer 15 Nov 2020 - 20 Jun 2021 Pink Tech Design Gurgaon, Haryana • Experience working with Computer Vision and used GDAL and Rasterio libraries to create satellite tif images suitable for deep learning models in the satellite imagery field. • Worked closely on Computer Vision models using CNN, YOLOV5, UNET, Mask R-CNN, GAN, Autoencoder and NLP models using RNN LSTM, Transformer. • Developed API’s of trained models using Django Rest Framework, integrated with React and Node applications, and deployed on the cloud. PROJECTS Advanced Virtual Assistant Platform • Developed an AI-based real-time virtual assistant with subscription payment integration, offering faster responses than typical mobile ChatGPT versions, by integrating chatbots and voice assistants for enhanced customer support. • Implemented features for seamless embedding of chatbots and voice assistants into websites using HTML, simplifying user integration. • Utilized external and custom local Large Language Models (LLMs) by integrating them with a LangChain agent to create tailored custom assistants. • Incorporated diverse knowledge sources to expand custom assistant capabilities, enabling tasks like appointment booking, web crawling, and lead generation. • Empowered businesses to deliver personalized and interactive experiences for website visitors, improving engagement and user satisfaction. • Positioned the platform as an advanced solution to traditional customer support, offering intelligent, conversational AI for websites. • Enabled more natural, context-aware, and meaningful user conversations, surpassing traditional rule-based chatbots. Real-Time Communication Analysis Tool • Developed a real-time call analysis tool using the MediaStream Recording API for live audio capture from browsers. • Employed WebSocket for server connection to facilitate real-time speaker diarization. • Integrated Whisper OpenAI STT (Speech-to-Text) for audio transcription. • Utilized Pyannote for speaker embedding to differentiate speakers during calls. Smart City Traffic Flow Optimization System • Designed and developed a system to analyze real-time video feeds from urban traffic cameras for optimizing traffic light schedules and reducing congestion. • Implemented object detection (vehicles, pedestrians) and tracking algorithms using OpenCV and a deep learning model (e.g., YOLOv5) to monitor traffic density and flow patterns. • Developed a predictive model using time series analysis (e.g., ARIMA, LSTM) on historical traffic data to anticipate peak congestion periods. • Created a simulation environment to test and validate different traffic signal control strategies before deployment. • Engineered a backend service using Python (FastAPI) for data ingestion from cameras and a frontend dashboard using React for visualization of traffic states and system performance. AI-Powered E-commerce Product Recommendation Engine • Developed a personalized product recommendation system for an e-commerce platform to enhance user experience and increase sales. • Utilized collaborative filtering and content-based filtering techniques, processing user interaction data (views, purchases, ratings) and product descriptions (NLP techniques on text). • Implemented the system using PySpark for scalable processing of large datasets of user behavior and product catalogs. • Integrated the recommendation engine via APIs with the existing e-commerce frontend (MERN stack components) to display personalized suggestions in real-time. • Employed A/B testing methodologies to evaluate the effectiveness of different recommendation algorithms and continuously improve model performance. Automated Document Summarization and Analysis for Financial Reports • Created a tool to automatically summarize lengthy financial reports and extract key insights using Natural Language Processing and Large Language Models. • Leveraged Hugging Face Transformers for abstractive and extractive summarization tasks on unstructured text data from PDF and TXT files. • Incorporated named entity recognition (NER) to identify and categorize key financial entities, terms, and figures within documents. • Developed a Langchain-based workflow to orchestrate data ingestion, pre-processing, model inference, and structured output generation (e.g., JSON with key findings). • Built a simple web interface using Streamlit for users to upload documents and view summaries and extracted information. EDUCATION Bachelor of Technology (IT) August 2016 - May 2020 Panipat Institute of Engineering and Technology TRAINING Completed Training On Machine Learning 7 May 2019 - 3 August 2019 Aptron Solution Private Limited Noida, UP • Implemented a variety of exploratory data analysis methods, including data collection, data cleaning, data preparation, data processing, feature engineering, and hypothesis testing utilizing Statistical Tools and created Machine Learning models. CERTIFICATIONS • IBM Applied AI Professional Course - Certified by EDX • IBM AI Engineering Professional Course - Certified by Coursera • IBM DevOps and Software Engineering Professional Certificate - Certified by Coursera • IBM Full Stack Cloud Developer Professional Course - Certified by EDX • Google Cloud Platform Professional Course - Certified by Coursera SKILLS Programming/Frameworks AI/ML Tools/Platforms Python Machine Learning Docker Django Rest API Deep Learning Kubernetes FastAPI LLM GCP JavaScript Hugging Face Transformers PySpark React Native Langchain Apache Kafka MERN Stack Computer Vision Apache Flink Node.js Natural Language Processing Elastic Search SQL (MySQL) Diffusion Neo4j NoSQL Prompt Engineering IBM Cloud HTML/CSS Time Series Forecasting Terraform Bootstrap 5 Bayesian Statistics Git Redux Tensorflow Agile Development PyTorch Keras Big Data Analysis Scikit-Learn Transfer Learning Full Stack Web Development Streamlit Optimization create portfolio based on my resume data - Initial Deployment
-
43 kB
complete it - Follow Up Deployment
-
388 Bytes
initial commit