Sadkirat Singh

WORK HISTORY

EDUCATION

ABOUT SADKIRAT SINGH

I am a software engineer bringing 3+ years of experience in scalable backend development. Having proven my ability in 0-to-1 product development, I am currently focused on mastering and overcoming challenges of 1-to-10 phase. I have successfully enabled clients to automate their processes using Applied AI Currently, leading engineering team of six people focused on delivering Applied AI solutions and designing algorithms to improve NLU I spearheaded 0->1 product development of kray.ai to enhance product discovery on e-commerce stores using AI search engines and chatbots Orchestrating requirements analysis and ensuring TDD to maintain high developer and stakeholder confidence and reduce time to UAT Enabling in-house sales through rapid prototyping, building POCs and technical training sessions to bridge communication between sales and engineering teams Here are few solutions we\'ve delivered for clients in field of Applied AI: 1. Installed Kray.ai services on e-commerce store of a US-based automotive company resulting in 74% improvement in product discovery, 27% reduction in exit rate and 66% increase in search results speed 2. Devised AI chatbots to enable Q/A on their data corpus with 85% accuracy 3. Architected document writing systems for legal and healthcare clients to reduce production time by 30%. Highlighting key efforts in NLP/NLU: 1. Engineered parse-tree service from scratch for proprietary NLP pipeline to improve NLU by 20% through syntactic parsing. Enhanced tokenization service to perform content-aware tokenization improving accuracy by a further 7%. Revamped legacy NER service from python2 to python3 to increase collaboration by 3x and enable namespaces and batch training. Trained AI-NLP model on 200+ historical sales proposals to recommend content snippets relevant to JD to enable in-house sales team to cut proposal creation time by 50%. I have also focused on R&D in space of Symbolic AI, Explainable AI and combining them with conventional methods of performing ML/DL. Shedding light on key efforts here: 1. Search Engine of kray.ai uses knowledge graphs(KG) and NLP. Augmented food safety evaluation service with ontology of 1M ingredient concepts extracted from DBPedia to enable identification of ingredients and their attributes inside ingredient strings. Empowered location extractor service with KG of 2M Geonames locations to detect and validate addresses present on web pages. Used various tools and frameworks to work with ontologies and KG like wikibase, protege, Neo4J, DBPedia, OWL, SPARQL, Apache JENA etc.