Most PMs use AI. I research it — and ship products with it.
6+ years in product across Telecom and B2B Enterprise.
3 NLP & GenAI research papers under review at Q1 journals.
Now building AI products full-time.
My career has moved in two deliberate phases. The first was product execution — shipping B2B software at Samsung and Sprint, working across the largest US carriers, learning how real products get built inside complex organizations.
The second was a conscious choice: go deeper on AI than most practitioners do. Original NLP research, now under review at Q1 journals, alongside hands-on product consulting. That combination — research instincts applied to real product problems — is what I bring to every engagement.
Currently open to roles in
Canada · Australia · Ireland · Netherlands · Germany · UAE · Singapore · USA · India
End-to-end product thinking — from original user research and PRDs to live prototypes and real sprint execution. Every project shows the full PM arc, not just the output.
Wrote the Working Agreement that kept a distributed 13-person team aligned across 3 sprints. MVP shipped in 6 weeks. Live product, zero scope creep.
A 17-page technical PRD for a multi-tenant RAG support system — dual-model routing, 9-point failure analysis, and token cost architecture. A product design exercise, not a shipped product.
Designed the guardrail that makes AI tutoring honest: it can never write the answer. Projected 2.5× lift in assignment pass rates.
33 surveys revealed the real reason courses get abandoned — not length, not boredom. Certificate credibility. 61% never finished because the credential wasn't worth the effort.
Made the product call that matters most in health AI: draw the line between insight and diagnosis. Designed for ≥25% early detection confirmation with ≤15% false positive rate.
45 million Americans are rejected by models that measure the wrong things. Designed the alternative-data architecture to fix it: +25–35% approval lift, same default rate.
Original research contributions examining how language models understand — and fail to understand — human language. Three works submitted to peer-reviewed Q1 journals following a year of independent research.
Investigates whether state-of-the-art Transformer models truly understand language systematically — or merely memorize statistical patterns. Demonstrates that scaling alone does not resolve the gap between statistical fluency and genuine compositional reasoning.
Proposes CBAE — a capsule network model that replaces attention mechanisms for aspect extraction in review text. Achieves superior F1 scores over ABAE and all baselines on the Citysearch benchmark.
Introduces OntoQA — a domain-agnostic question-answering platform using LTAG/DUDES linguistic formalism and greedy parsing to translate natural language into precise SPARQL queries against structured knowledge bases.
Led end-to-end product development of an AI chatbot platform integrated across EdTech and eCommerce client websites — owning full PRD, user journey mapping, conversational flow design, and KPI framework from 0 to 1. Defined product requirements for LLM-based conversational flows, collaborating with engineering on integration architecture, prompt design, and UAT. Drove independent NLP and GenAI research in parallel — 3 original contributions examining transformer-based text understanding and semantic query resolution, submitted to peer-reviewed Q1 journals.
Owned the product roadmap for a B2B network support platform serving Verizon, AT&T, and T-Mobile across 4G LTE and 5G. Managed a 100+ story backlog — delivering a 30% CSAT increase and 20% faster time-to-market.
Shaped roadmaps across 3 consecutive development sprints from market research and stakeholder inputs. Authored user stories achieving a 95% first-pass validation rate — eliminating a full regression cycle per release.
28-week internship across O&M, Switch Planning, and IP Transmission — supporting FTTH, IMS, GIS, and B2B infrastructure projects under the Head of Planning.
Aman brought real structure to a complex product environment. He coordinated across teams, managed competing priorities, and consistently delivered accurate, timely results under pressure. A dependable contributor who takes ownership end to end.
Aman took a genuinely complex product problem — with minimal direction — and delivered an analysis that directly shaped our roadmap decisions. His methodical approach and independent thinking made a real difference to the team and to our customers.
Aman has excellent analytical abilities and a genuine drive to master the deeper layers of a subject. His work on frequency extraction using Cepstral Analysis was an exceptionally high-quality contribution — efficient, well-reasoned, and technically rigorous.
Aman is adept at managing intricate tasks independently. His problem-solving ability is distinctive — he approaches challenges in innovative ways and brings resourcefulness that is rare at his career stage. One of the brightest people I have worked with.
Aman consistently collaborated with peers to introduce better methods and fresh ideas. He has an excellent sense of perception regarding technical concepts, and his supportive, positive attitude creates a genuinely productive environment for everyone around him.
Aman is a reliable professional who translates intellectual rigour into tangible results. His ability to organise complex ideas clearly and communicate them effectively makes him a genuine asset. He has the maturity and competence to perform at a high level in demanding environments.
If you're building an AI product and your PM can't engage with the research behind it — that's a solvable problem. Let's talk.