Divya Goyal

Engineering leader with 11+ years of experience driving AI innovation across ML, NLP, Generative AI, and LLMs. Have led the end-to-end development of SirionAE and other intelligent systems—from research to scalable deployment—using PyTorch, TensorFlow, Hugging Face, and Kubernetes across AWS, Azure, and GCP. Expert in building production-ready GenAI solutions, LLM-based agents, and multi-modal conversational platforms (MCP), with strong focus on MLOps, cloud-native infrastructure, and business impact.

Core Competence:
Machine Learning & Deep Learning | AI Strategy & Roadmap Development | Generative AI (GPT, GANs, Multi Agent
framework) | Data Science & Analytics | Cross-functional Leadership | Product Development & Innovation | Scalable System
Architecture | Research & Development (R&D) | AI Ethics & Governance | Talent Development & Mentorship | Strategic
Partnership Management | Agile & Lean Methodologies

• Programming Languages: Python, Keras, Tensorflow, C/C++
• Frameworks & Tools: TensorFlow, PyTorch, Keras, Scikit-learn, Apache Spark, Docker, Kubernetes
• Databases & Cloud: AWS, GCP, Azure, MongoDB, PostgreSQL, NoSQL, SQL
• ML Platforms: Google AI Platform, Amazon SageMaker, Azure ML
• Data Engineering Tools: Hadoop, Spark, Kafka, Apache Pulsar, Azure Databricks
• Machine Learning: Predictive Modeling, Classification, Regression, Reinforcement Learning, Deep Learning, Transfer Learning, Speech Recognition, GANs
• Gen AI / NLP/ NLU Techniques: Text Mining, BERT, Prompt Engineering, Azure OpenAI GPT 3.4/4, Lllama-3, Langchain

SIRIONLABS

• Developed LLM based systems which has talk to document and talk to corpus being the main features.
• Developed micro architecture for extraction and classification of data using Kafka, Mongodb and Rest architecture.
• Delivered end to end scalable software feature.
• Built features and optimized (hyper-parameter tuning) the state of the art machine learning and deep learning techniques on CPU/GPU environments.
• Customized Spacy's NER for information extraction and applied transfer learning techniques using pre-trained word embeddings like BERT for classification.
• Automated the data science platform by building and managing data pipelines and extensions that bridges NoSQL-databases, APIs, Machine Learning Engine and UIs.
• Containerized(Docker) and deployed the re-producable machine learning model into production.
• Business problems solved: Duplicate documents using text similarity, Clause classification, Information Extraction from Contracts.

BANK OF AMERICA

• Developed system for calculation of risk and P&L useful for traders.

ECLERX

• Robotics Process Automation - Developed a robotics process automation application to automate the trading processes for an investment bank in order to reduce human efforts. Also prepared technical design of various processes and application.
• Work Force Management Software - Implemented a system for effective scheduling of work force, which involves creation as well as the design. Developed a software for providing breaks which reduced the manpower by 40 FTEs. Evaluated different algorithms for the optimized and cost-effective solution to minimize operational costs.
• Warehouse Inventory Management System - Includes the life-cycle of the product from inventory to the technician, then to customer and back to inventory. Developed the product from scratch that facilitated an organization working in CTS domain. Prepared Payroll processing module for technicians by linking it with the inventory.
• ResqRangers Android App - Developed an android app from scratch in the area of Animal Crisis Management. It includes the interface for reporters, volunteers, NGOs, Para Professionals, ambulance.
• Speech to Text - Standalone project, includes research and enhancement to the open source software by CMU. Designed and implemented a tool for the business requirement. Optimized the existing model by dealing with the Language, Acoustic and Grammar models.

DRDO

• Created a web based application for managing inventory of medicines using SQL and PHP.