Feb 2023 – May 2025 · 2 yrs 3 mos · Bengaluru, Karnataka, India · Hybrid

• Worked on Symptom Checker Chatbot of Apollo 24/7 app which handles around 45,000 chats per day. This enquires the patients about their symptoms and recommends doctor consultation, lab test, self-care advice, possible diseases with around 70% accuracy. Worked on a Lab Recommendation service, added questions based on patient’s medical records in chatbot, implemented A/B testing in production using GrowthBook’s API.
• Worked on building a Doctor’s Module that shows patient’s recent chat data and past medical, surgical, family history, ongoing medications, etc. to the doctor when the patient comes for consultation. This improves consultation experience for both doctor and patient.
• Built a Clinical Knowledge Graph from Apollo’s patient data, casesheets, and medical dictionaries like UMLS. Defined the database schema in BigQuery (SQL) and Neo4j (Graph DB). Created data ingestion pipelines in Python, enabling data-driven recommendations and analysis.
• Built a customizable recommendation service leveraging the Clinical Knowledge Graph, generating lab, medication, and doctor specialty suggestions from user’s past history, diseases, purchase history, etc. This service enhanced user engagement by offering tailored healthcare recommendations, improving conversion rates by 5% for lab tests.

Jun 2021 – Feb 2023 · 1 yr 8 mos · Bengaluru, Karnataka, India · Remote

• Worked on adding AI features like face 2D and face 3D filters, background segmentation, facial expression detection, style transfer etc. in ShareChat camera.
• Worked on unwanted object removal from image. After benchmarking of different computer vision algorithms and deep learning based models, picked up Hifill Model and reduced model size by 80% with acceptable performance to deploy on mobile devices.
• Optimized TensorFlow model loading for GPU inference, reducing load times by 50%, using model quantization, pruning, tflite conversion, model preloading.
• Wrote a cross platform library in C++ with common interface for Android, iOS, and MacOS for running AI pipeline with support for adding multiple AI modules. Integrated this library to ShareChat’s in-house graphics library ShutterLib.
• In Graphics library, moved 3D filter and AI asset loading codes to background thread to resolve UI freeze, when switching between two camera filters.
• Made an android app for our AI team where team members can test developed features before integrating to ShareChat camera.