Gourav Sinha
AI Engineer (Technical Lead, Applied Research) @ Siemens | GenAI & Agentic Systems | Neuro Symbolic AI | EPAT certified algo trader
Pune District
11+ years of work experience
Sep 20232 yrs 11 mosPresent
Pune District, Maharashtra, India
Working on 'Building X' product
Domain : Smart Buildings and infrastructure.
1. Added features like precise image and link rendering in Building X chat assistant with the help of intelligent scraping of html user docs.
2. Built a customized transformer from scratch using KAN to generate accurate iot device descriptions from iot device abbreviations.
3. Built and deployed chatbot using agentic rag with langraph and ensemble retriever(faiss and bm25) on around 9k document chunks with ollama service leveraging mistral instruct . Used mixbread embedding model to generate faiss vectorstores. Used customized strategy to chunk images and text from html web pages.
4. Trained Marian-nmt architecture from scratch to translate German technical documents.
5. Has used and experimented Lora and Quantization with AWQ and GPTQ quantization techniques.
6. Has built encoder decoder translation architecture with bert and gpt 3 respectively. Used it to transform technical iot device short descriptions to detailed technical specifications. Bleu score for this experiment was around 0.7
7. Have implemented customized clustering algorithm with umap for dimensionality reduction and hdbscan on embeddings generated from customized encoders that were trained on domain specific technical documents.
Aug 20212 yrs 2 mosSep 2023
Pune District, Maharashtra, India
NATURAL LANGUAGE PROCESSING:
1. Customised NER model using Longformers with pytorch to extract company services keywords from ~12k companies.
2. Built keywords clusters using Bertopic to eventually cluster similar companies.
3. Carried out text Augmentation and synthetic text generation using GPT-2 to generate and collate accurate datasets for training the news article classification model to identify rare type of news articles.
4. As part of domain specific internal NLP research Leveraged KL divergence to create a keyword clustering methodology based on probabilistic distribution of text data.
5. Developed end to end NLP pipeline to perform timely keyword extractions and perform retraining on negative samples based on SME's feedback using Dockers, AWS ECS and S3.
6. Did Backend development with PostgreSql , MongoDB , Elastic search etc to build fast NLP model APIs.
FINANCE AI:
1. Customer segmentation based on different parameters like equity investments, AUM and other such engagements.
2. Recommending relationship mangers based on leads and opportunities in CRMs.
3. Lead extraction from news sources using NLP.
Experience in using LLMs /Langchain/Prompt engineering.
Experience in building customized GPTs from scratch.
Performed Lead roles , carried out code reviews on Git and mentored juniors.
Apr 20215 mosAug 2021
Feb 20156 yrs 2 mosMar 2021
India
1. Built predictive models to predict the probability of return of a particular product.
2. Built and deployed an end to end framework that monitors background jobs and predicts the action that needs to be taken in case if the job fails.
3. Built a ticket triaging model that predicts assignment group based on short and long description of tickets.
4. Led a team that worked on frontend and backend automation.
5. Built customised knowledge graph using django rest api and vanilla javascript to represent more than 5000 workflows and it's execution. Deployed the application on AWS ECS
Gourav’s AI Stack
Education
PGP, Artificial Intelligence
Texas McCombs School of Business
20201 yr2021
Bachelor of Engineering (B.E.), Computer Software Engineering
Rajiv Gandhi Prodyogiki Vishwavidyalaya
20104 yrs2014
Join Gourav and 10k+ other professionals on
Tal is the app where bosses meet engineers & builders directly, without middlemen
Download Tal Boss