Ashish's Profound AI Rep
Ashish's view on problem-finding shows how he approaches every technical decision from the business lever down.
Ashish Kumar
Edges
Building Enterprise AI Pipelines
Ashish designs and ships end-to-end AI systems for enterprise clients, owning decisions from architecture selection through cloud deployment. He has done this across RAG knowledge bases, multimodal document processing, and real-time message classification pipelines, working with clients in consulting, supply chain, and financial services. His approach starts with explicit trade-off analysis: evaluating database options, model choices, and deployment platforms before writing production code. That discipline shows up in the Dalberg RAG system, where he identified vision-language models as the right tool for PPT and image extraction, and in the WhatsApp automation, where he redesigned the infrastructure with AWS Lambda to handle order-of-magnitude scaling. The result is systems that hold at scale and produce outcomes clients can measure.
Spotted in 2 Stories
Business Lever Identification
Ashish approaches technical problems by first finding the smallest measurable input that actually drives the business outcome, then building toward that lever. At Unacademy, he traced the revenue problem through multiple layers of the sales funnel until he landed on talk time per BD as the controllable input metric, then designed the auto-dialer specifically to move that number. At Tailored AI, he frames client engagements around the business problem first, using that framing to make architecture decisions. This instinct comes partly from his time as a product manager and partly from his operations background, both of which trained him to read systems before touching them. For engineering teams that need someone who can translate a business problem into a technical brief without losing fidelity in either direction, that combination is the practical value.
Spotted in 2 Stories
U
Cross-Functional Product Engineering
Ashish has worked as a software engineer, a product manager, and an operations manager, and he carries all three frames into every role he holds. At Unacademy, he was a PM who ran stakeholder interviews at every level of the sales org, mapped the metric hierarchy, and designed a feature with a clear input-output logic. At Tailored AI, he is an engineer who scopes client problems, evaluates trade-offs, and owns the full build from architecture to deployment. That range means he can hold the technical and business sides of a problem simultaneously, without needing a handoff between functions to translate between them. For early-stage or lean AI teams where engineers are expected to understand the product context, that combination reduces coordination overhead and speeds up the path from problem to shipped solution.
Spotted in 2 Stories
Ashish Kumar
About Me
Ashish Kumar is an AI engineer and former product manager with experience spanning software engineering, product, and operations. He holds a BTech from IIT Bombay and currently builds enterprise-grade generative AI solutions at Tailored AI, working across RAG pipelines, LLMs, OCR, and multimodal systems for clients in consulting, supply chain, and financial services.
At Tailored AI, he architects and ships end-to-end AI systems for enterprise clients, owning decisions from model selection and infrastructure design through to cloud deployment.
For Dalberg, the nonprofit arm of BCG, he built a RAG knowledge base that achieved 95% recall on gold-standard test data, processing PDFs, DOCX, and PPTs using vision-language models for accurate multimodal extraction.
At Unacademy, he shipped an auto-dialer feature for the inside sales CRM that drove a 5 to 7% jump in call volume by eliminating manual lead navigation.
Earlier, he built a WhatsApp-to-system automation for a supply chain client, scaling from 200 daily messages to over 6,000 using AWS Lambda without system degradation.
Ashish Kumar
Half the problem is already solved the moment you know what you're actually trying to move.”
Ashish Kumar
On finding the right problem