Profile
Seven years at Deloitte delivering data and reporting systems for US public programmes, followed by two years building applied AI systems.
My work spans requirements, architecture, trade-offs and evaluation: deciding what a system should do, how it should behave when evidence is weak, and how to prove whether it works.
NeuralHue AI Limited
Applied AI for regulated industries. Three engagements, one product.
Founder · 2024 - present
FirmMemory
A governed retrieval layer over a law firm's matter archive, built for the question keyword search cannot answer: have we advised on something like this, and what came up. Paragraph level citation, deterministic gap detection, and refusal when the evidence is thin. Evaluated with a 30 query adversarial kit across six axes, with measurement noise characterised at plus or minus one query.
Semantic talent intelligence
Production search platform for a legal talent intelligence firm. Separate vectors for experience, matters and publications, fused only at ranking time, so a search can distinguish delivery experience from thought leadership. Client owned and extended beyond scope.
ESG disclosure assurance
Scoped an assurance engine for Indian listed companies. Source authority stamped at acquisition rather than inferred, making the boundary between assertable findings and surfaced third party claims structural rather than a policy.
Learnt
The governed layer, meaning provenance, coverage and refusal, is the part clients should pay for. The one engagement where it was scoped out is the only one that produced nothing.
Deloitte Consulting
US state and federal programmes. Eligibility, public health and welfare. Data, reporting and analytics delivery. High stakes data quality and statutory reporting.
Consultant → Senior Consultant · Strategy & Analytics · 2016–2023
Techno functional lead between US client leadership and India delivery teams, and functional point of contact for federal reporting policy. Promoted to Senior Consultant in 2019.
State COVID-19 vaccination programme
Built a data quality and integrity framework on Azure Synapse during a live national rollout, cutting data errors roughly 30% and making patient profiling reliable enough to act on.
Federal and state statutory reporting
Designed and delivered the reporting infrastructure, taking data compilation and statutory submission from several weeks to two days.
Healthcare and welfare legislation
Delivered analysis and reporting to US state government leadership that informed legislation, in a programme where a reporting error changes an individual's benefit entitlement.
Post merger data integrity, US healthcare
Designed the framework that identified and remediated data gaps across a merger, keeping records continuous through migration.
Telecom reporting platform
Built cloud pipelines on AWS using Python, Redshift and Apache Superset, then led an Azure Synapse migration that cut reporting latency around 40%.
Pursuits and practice
Contributed to sales pursuits, proof of concept builds and RFP responses. Mentored junior consultants.
Learnt
Nobody on a public programme asks for provenance, because a number you cannot trace back is not a number. That expectation is the one most AI systems have not caught up with.
Education
Executive Programme in Business Analytics in the Age of Generative AI, London Business School, 2025
MBA, Symbiosis Institute of International Business, 2016
B.Tech Computer Science, SRM Institute of Science & Technology, 2013