Ayush Mittal — Founder
I’m a data scientist specialising in algorithm design for applied ML systems: ranking, retrieval, representation learning, probabilistic models, optimisation and LLM-based systems. My focus is on building new algorithms and model architectures tailored to the problem. For 10+ years I’ve worked on: - Ranking and recommendation systems - Representation learning and embeddings - Probabilistic modeling and graphical models - Optimisation and decision algorithms - Generative models and large language models (LLMs) - Data/intelligence systems at scale (logs, events, tabular) I care about formulating problems precisely, choosing the right objective functions and inductive biases, and understanding the failure modes of models in real environments. Some recurring technical themes in my work: • Ranking, retrieval and recommendations - Multi-stage retrieval and learning-to-rank pipelines over 300M+ entities - Wide & deep models, sequence models (LSTMs), and embedding-based architectures for feed ranking and candidate generation - Multi-objective optimisation for recommender systems (engagement, retention, risk, business constraints) • Document understanding, structure induction and knowledge graphs - Document-to-knowledge-graph conversion pipelines using beam search, constraint satisfaction and entity-linking - Normalisation and alignment of heterogeneous clinical / operational documents - Information extraction for adverse events, operational signals and site intelligence • Probabilistic and decision models - Probabilistic graphical models for drug adherence and behavioural patterns - Risk scoring models combining structured, temporal and graph signals - Throughput and scheduling optimisation for manufacturing using queueing models, simulation and constrained optimisation • LLMs, generative models and code/data automation - Training and adapting domain-specific LLMs and representation models for search, Q&A and decision support over specialised corpora - Retrieval-augmented generation with custom retrieval layers, ranking and evaluation loops - Code generation models for SAS/Python/R targeting ETL, data validation and analytics, including safety filters and evaluation harnesses • Detection and quality systems - NSFW and duplicate content detection via CNNs and representation learning - Risk-based data issue detection combining statistical profiling, anomaly detection and supervised models - Automated data validation rule learning and monitoring for large-scale data pipelines
Stackforce AI infers this person is a Healthcare AI expert with strong capabilities in algorithm design and data science.
Location: Bengaluru, Karnataka, India
Experience: 10 yrs
Skills
- Algorithm Design
- Applied Ai
- Data Science
- Machine Learning
- Data Engineering
- Computer Vision
Career Highlights
- Expert in algorithm design for ML systems
- Proven track record in healthcare AI solutions
- Led teams to enhance user engagement significantly
Work Experience
ZetaGrade
Founder & Director (1 yr 9 mos)
Lifio.ai
Co-Founder & CTO (3 yrs 9 mos)
Fresh Gravity
Senior Manager (4 yrs 1 mo)
Data Science Advisor (1 yr 9 mos)
Manager (5 mos)
Sr. Consultant (10 mos)
Stealth Startup
Founder (4 yrs 4 mos)
ShareChat
Lead Data Scientist (1 yr 4 mos)
Visa
Senior Software Engineer (10 mos)
VisageMap Inc.
Intern (2 mos)
Indian Institute of Technology, Kanpur
Teaching Assistant (4 mos)
Microsoft Research
Research Intern (2 mos)
Education
Master’s Degree at Indian Institute of Technology, Kanpur
Bachelor of Technology (BTech) at Indian Institute of Technology, Kanpur
at St. Paul's Higher Secondry School Indore
