• Prompt Engine & RBAC based Prompt management for organisation wide R&DS applications (PaaS). • De-centralised helm based shippable product of same.• Istio service mesh for monitoring & destination based rate-limitng/retries etc. • Single/Multi-tenant nofication service for OpenAI prompt engine (PaaS) • Search & Analytics - Elastic Search and FastAPI with DB sync (logstash/filebeat ecosystem)• Multi-node Elastic search/Kibana/logstash/filebeat deployment and scaling• Identity management and backend API authentication (OIDC)• Scalable Product level authentication with ingress-controller based routing• Event-Driven data pipeline with Apache airflow, celery workers & Logstash• Author reusable boiler-plate code & components

• Design & implement provisioning of big-data jobs & resource utilisation tracker across the platform• Design & implement python/pyspark partner app versioning feature• Author Inhouse-utilities, boiler plate codes & libraries• Implement Multi-tenant shared wallet & DB auth for partner apps

• Built a DNN model for heartbeat categorization using Keras and Tensorflow APIs.• Implemented configuration-driven data pipeline builder improving code reusability.• PoC on Apache Kafka based system for rhythm analysis and arrhythmia classification on bulk data

• Improved data preprocessing pipeline for model inferences, making the end-to-end process more than 75% faster by code optimization and parallel processing.• Extracting signal from JPG/PNG images using open-CV2, preserving amplitude and time information• Plot API for generating medical-grade ECG plot (Independent of the acquisition system used)

• PoC on medically relevant keyword extraction software with NER model built using Stanford CoreNLP.• SQL queries for generating analytics reports on user engagement and their interactions.

• Extraction of keyword, entity and category from user-generated text using IBM, Microsoft and Google NLP API for curating user interests.