Kanish Tyagi — I Build What Users Need

The Arc

Most people pick a lane.
I picked the whole highway.

Every role taught me something the next one needed. QA taught me how products fail. Data Science taught me how to measure what matters. Product taught me how to decide what to build next.

APR 2022 – NOV 2024

QA Engineer

Wingman Partners · Noida, India

Owned quality across 4 product modules. Built Selenium automation suites that cut manual regression by 40%. Reduced defect leakage by 30% through structured root-cause analysis. Worked cross-functionally with PMs and devs — learned firsthand how product decisions ripple into real user pain.

Selenium Jira API Testing Confluence Regression Analysis

OCT 2021 – MAR 2022

Software Engineer Intern

Wingman Partners · Noida, India

First real engineering work. Contributed to full-stack web applications — HTML/CSS front-end, API integrations, database queries. Learned how software actually gets built before learning how to break it.

HTML/CSS REST APIs SQL Full-Stack

JAN 2025 – DEC 2026

M.S. Data Science

University of Texas at Arlington

Studying the math behind the decisions: ML, optimization, statistical inference. Built production-grade ML models (87% accuracy malware detection, employee attrition prediction). AWS GenAI Hackathon 2nd runner-up. Speaker at UTA Business Analytics Symposium 2026.

Python ML / Deep Learning Convex Optimization NLP LLMs

JAN 2026 – MAR 2026

Product Strategy Contributor · ConsenTerra AI

Early-Stage AI Startup · FoundersBoost NYC Spring 2026

Designed B2B pricing framework (3-tier model: $99/$249/$499), built onboarding questionnaires, positioned GTM strategy for enterprise buyers. Worked directly with the founder on product-market fit. Real startup, real accelerator — not a class project.

B2B SaaS Pricing GTM Strategy User Onboarding Product-Market Fit

JUN 2026 – AUG 2026

AI Business Analyst Intern

Triple-S Steel Supply LLC · Houston, TX

Working under the Director of Digital Transformation to design and ship AI-powered automation workflows. Built an end-to-end intelligent document processing pipeline that cut manual processing time for operational teams significantly. Invited to extend through August 15 after initial engagement due to project impact. Currently architecting a second project: an AI orchestration layer connecting multiple enterprise data agents into a single unified interface.

Microsoft Azure AI Copilot Studio Power Automate TypeScript AI Orchestration Process Automation

What I Actually Do

Not buzzwords. Proof.

Every skill here has a story behind it — a project shipped, a problem solved, a metric moved. Click any card to see what I've actually done, not just what I claim to know.

🎯

Product Strategy & Management

I've priced SaaS, designed onboarding, and positioned products for enterprise buyers.

Built 3-tier B2B pricing model for ConsenTerra AI (Starter $99 → Growth $499/mo)
Designed onboarding questionnaires that reduced time-to-value for new users
Wrote GTM positioning documents for enterprise AI compliance buyers
Evaluated 20+ job postings with structured go/no-go frameworks — PM thinking applied to my own search
Sent cold emails to founders with PM-style friction-point critiques of their products

Jira Figma Notion Roadmapping PRDs User Research

📊

Data Science & Analytics

I build models that work, not just notebooks that pass.

87% accuracy Random Forest model predicting employee attrition — deployed on Streamlit
Built malware detection system using decision trees on IoT network data (EE5364 course project)
Analyzed 60,907 Stack Overflow responses to quantify GenAI's ROI on developer productivity
Customer segmentation using K-Means clustering for targeted marketing
Airbnb price EDA — feature engineering across 30+ variables

Python Pandas Scikit-learn TensorFlow SQL Tableau Streamlit

🤖

AI, GenAI & Open Source

I don't just use AI tools — I contribute to the ones other people use.

7 merged PRs to Microsoft's agent-governance-toolkit — CodeQL, CSP headers, Colab notebooks, docs
Built Adaptive AI Study Tutor using Multi-Armed Bandit RL + Claude API
Wrote 5 technical blog posts on AI governance published to Microsoft's community docs
Built policy YAML validation CLI for AI governance workflows (open PR)
AWS GenAI Hackathon 2nd runner-up across Texas Universities

LLMs Claude API GitHub Actions TypeScript Python CodeQL Google Colab

🔍

Quality Engineering & Testing

3 years of finding the bugs before users do.

Owned QA for 4 product modules across 10+ RESTful API endpoints at Wingman Partners
Built Selenium + Python automation suites — reduced manual regression effort by 40%
Achieved 30% defect reduction through structured bug analysis and prioritization
Documented processes in Confluence — maintained test documentation for the entire QA team
ISTQB Certified Foundation Level — formal testing methodology

Selenium Postman Jira Confluence CI/CD REST APIs ISTQB

Proof of Work

Talk is cheap.
Here's what I shipped.

Product Strategy

ConsenTerra AI — B2B Pricing & GTM

Designed the entire pricing architecture and go-to-market positioning for an early-stage AI compliance startup. Built 3-tier model, onboarding flows, and enterprise buyer personas. Startup accepted into FoundersBoost NYC Spring 2026 accelerator.

Impact: Pricing framework adopted by founder. GTM positioning used in accelerator pitch. Founder publicly credited work on LinkedIn.

Open Source · Microsoft

Agent Governance Toolkit Contributions

7 merged PRs to microsoft/agent-governance-toolkit. Added CodeQL security analysis for TypeScript, Content-Security-Policy headers, three Google Colab onboarding notebooks, five Dev.to blog posts, and comprehensive Python docstrings.

Impact: Direct contributions to Microsoft's open-source AI governance infrastructure used by enterprise teams.

GitHub → Blog Posts →

AI / RL

Adaptive AI Study Tutor

Built an intelligent tutoring system using UCB1 Multi-Armed Bandit reinforcement learning + Claude API. The system adapts question difficulty in real-time based on student performance. Deployed on Streamlit.

Impact: Personalized learning paths generated dynamically — not a static quiz, an actual adaptive system.

GitHub →

Data Science

Health Center Attrition Predictor

End-to-end ML pipeline predicting employee attrition using Random Forest. Feature engineering across 35 variables, hyperparameter tuning, interpretability analysis. Deployed as a live Streamlit dashboard.

Impact: 87% accuracy. Identified top attrition drivers: overtime, job satisfaction, distance from home.

Data Analysis

GenAI ROI — Stack Overflow 2024

Analyzed 60,907 developer survey responses to quantify the real productivity impact of GenAI tools. Segmented by experience level, language, and company size. Built interactive visualizations.

Impact: Data-driven framework showing where AI tools help most vs. where they don't — not hype, evidence.

ML / IoT Security

IoT Malware Detection System

Course project (EE5364) building decision tree classifiers using entropy and mutual information on CTU-IoT-Malware-Capture dataset. Full pipeline in Google Colab with feature selection and evaluation metrics.

Impact: Academic project demonstrating applied ML on real-world cybersecurity data — not toy datasets.

AI Automation · Internship

Intelligent Document Processing Pipeline — Triple-S Steel

Designed and built an end-to-end AI automation workflow for processing operational documents. Integrated Microsoft Copilot Studio, Power Automate, and Azure AI to eliminate manual data entry steps across the operations team. Delivered within first month of internship; extended engagement based on results.

Impact: Dramatically reduced manual processing time for operational workflows. Extended internship through Aug 15 due to project success.

AI Orchestration · Internship

Multi-Agent AI Orchestration Layer — Triple-S Steel

Architecting a unified AI interface that connects multiple disconnected Microsoft Fabric Data Agents into a single conversational layer. Scoped, designed, and currently building the orchestration logic to allow non-technical stakeholders to query across data sources via natural language.

Impact: Enables enterprise teams to access fragmented data intelligence through one interface — no SQL, no switching tools.

Newsletter · Content

The F-1 Brief — International Student Newsletter

Founded and run a newsletter on Beehiiv covering CPT/OPT timelines, visa strategy, and career navigation for F-1 international students in the US. Content built from personal experience — I missed job offers due to processing delays and turned that pain into a resource for others.

Impact: Growing subscriber base of international students navigating US work authorization. Companion to the OPT/F-1 Survival Toolkit on Gumroad.

Education

The foundation.

M.S. Data Science

University of Texas at Arlington

Jan 2025 – Dec 2026 · GPA 3.8

Machine Learning, Deep Learning, Convex Optimization, NLP, Statistical Inference. AWS GenAI Hackathon finalist. Speaker at Business Analytics Symposium 2026. CPT-eligible May 2026.

B.Tech Information Technology

UPTU Lucknow (IIMT College of Engineering)

Jul 2018 – Jul 2022

Foundation in computer science, algorithms, databases, and software engineering. National-level competitive athlete during undergrad — Discus Throw and Shotput.

Certifications & Recognition

Credentials that back it up.

ISTQB Certified Tester — Foundation Level

ISTQB

Google Data Analytics Professional Certificate

Google / Coursera

2nd Runner-Up — AWS GenAI Hackathon (Texas Universities)

AWS

Speaker — UTA Business Analytics Symposium 2026

UT Arlington

FoundersBoost NYC Spring 2026 — ConsenTerra AI

FoundersBoost

Microsoft Open Source Contributor — 7 Merged PRs

Microsoft

Beyond the Resume

I'm not just a candidate.

The best products are built by people who understand humans, not just systems. Here's what shapes how I think.

🥇

National-Level Athlete

Competed in Discus Throw and Shotput at the national level. Discipline, pain tolerance, and structured training — the same traits that make me grind through ambiguity in product work.

💪

Fitness & Nutrition Nerd

I track macros, program my own workouts, and geek out on exercise science. Consistency compounds — in the gym and in careers.

🧮

Numerology Enthusiast

I study patterns in numbers the way most people study horoscopes — except I actually find useful patterns. Unconventional? Sure. But curiosity has no dress code.

📦

Digital Product Builder

Launched the "OPT/F-1 Survival Toolkit" on Gumroad — a 10-section guide for international students navigating CPT/OPT. Built from personal experience of missing job offers due to processing delays.

📰

Newsletter Founder — The F-1 Brief

Running a Beehiiv newsletter helping F-1 international students navigate CPT/OPT, job timelines, and visa strategy in the US. Turned my own stressful experience into a weekly resource for others in the same boat.

I broke software for 3 years.
Now I build what users actually need.

Most people pick a lane.
I picked the whole highway.

Not buzzwords. Proof.

Talk is cheap.
Here's what I shipped.

The foundation.

Credentials that back it up.

I'm not just a candidate.

I sit at the intersection of
Product, Data, and AI.

I broke software for 3 years.Now I build what users actually need.

Most people pick a lane.I picked the whole highway.

Not buzzwords. Proof.

Talk is cheap.Here's what I shipped.

The foundation.

Credentials that back it up.

I'm not just a candidate.

I sit at the intersection ofProduct, Data, and AI.

I broke software for 3 years.
Now I build what users actually need.

Most people pick a lane.
I picked the whole highway.

Talk is cheap.
Here's what I shipped.

I sit at the intersection of
Product, Data, and AI.