Cross-Market Technical Indicator Backtesting System
Research system evaluating 23+ technical indicators across US and Indian equity markets.
Published on SSRN — 21,528 strategies tested
PythonQuant modelsBacktesting frameworks
Home
Krishna Harsha Tangirala
Dubai, UAE — Applied AI Engineer
M.S., Applied Artificial Intelligence · University of Wollongong in Dubai
I build intelligent systems that automate decisions.
AI engineer focused on building production systems across finance, risk, and automation.
4y 5mProfessional experienceAmazon → Uber, 2022–2026
19Projects shippedPublic + private repositories
2Research papers publishedSSRN · market-efficiency & signal evaluation
- Uber
- Amazon
- SSRN-published
- ISB
- Masters' Union
01Applied AI Engineer
02Systems Builder
03Quantitative Thinker
- Machine Learning & AI
- Data Science
- Financial Markets
- Quantitative Systems
- AI-Driven Decisions
- AI × Finance
- Applied Research
- Real-World Impact
- Production AI Systems
- Cross-Region Deployment
- Systems thinking over point solutions
- Data-driven, execution-first
- Built to scale, measured by impact
- Continuous iteration
Applied AI Engineer focused onIntelligent automation systemsFinancial and risk modelingAI-powered decision infrastructure
02
Experience
Three roles. Click any card for the full breakdown.
Team Leader, Global Account Takeover (GATO)
Uber · Mar 2026 — Jul 2026
- Built
- Led a $30M+ agentic AI initiative across US&C and EMEA
- Impact
- 62%+ more accurate fraud detection · ~40% lower operating costs
- Won
- CST Extra Award, Uber (2026)
Risk & Compliance Specialist II
Uber · Feb 2024 — Mar 2026
- Built
- Built UWFMS, an AI/ML fraud investigation system
- Impact
- 35% faster investigations · 20% higher QA accuracy · 80+ analysts trained
- Won
- Golden Ticket & Top Performer, Uber (2024)
Seller Partner Support Associate
Amazon · Feb 2022 — Feb 2024
- Built
- Managed supply chain and seller operations for North America
- Impact
- Built forecasting dashboards; strengthened financial controls
- Won
- Top Performer, Amazon (2023)
03
Projects
Five systems where the outcome was measured, not estimated.
02
Bonteous × ISB Consulting Engagement
Consulting engagement translating ambiguous business problems into structured analytical frameworks.
Delivered data-driven business insights for Bonteous
Business AnalyticsConsultingData-Driven Insights
UWFMS / Bliss Scout — AI Fraud Detection Platform
End-to-end AI system combining knowledge graphs with ensemble ML models for fraud detection. Built at Uber while serving as Risk & Compliance Specialist II.
35% faster investigations · 20% higher accuracy
PythonML modelsKnowledge Graphs
SMC/ICT Statistical Edge Research
Formalizes 44 Smart Money Concepts and ICT trading signals as explicit detectors and tests them against a composition-matched random-entry baseline on the S&P 500, not just a zero-return null.
Published on SSRN — a corrected re-analysis finds none beat random entry
Pine → PythonFDR correctionCalendar-time inference
DCF Valuation + Monte Carlo Simulation Engine
Financial modeling system for valuation and probabilistic forecasting.
Scenario-based valuation · portfolio risk estimation
PythonFinance modelsSimulation systems
Also on record
04
Certifications
Every credential opens to what's on record — the certificate itself, where one has been issued.
B.Tech, Computer Science & Engineering
GITAM University · 2017 — 2021
First Class
Postgraduate Diploma in Operations Management (PGDIOM)
Indira Gandhi National Open University · Jan 2023 — Aug 2024
4.04 / B / 68.26% (First Division)
Advanced Management Programme in Business Analytics (AMPBA)
Indian School of Business (ISB) · Jul 2024 — Feb 2026
Successfully completed
PGP in Capital Markets & Trading
Masters' Union · Jul 2025 — Oct 2026
Overall CGPA 3.46 · Graduation status: awaited
M.S., Applied Artificial Intelligence
University of Wollongong in Dubai · Sep 2026 — Present
Currently enrolled
05
Awards & Recognition
- Led a $30M+ AI fraud prevention initiative
- Improved fraud detection by 62%+
- Reduced operational costs by ~40%
- Reduced investigation time by 35%
- Trained 80+ global analysts
06
Skills
Stated, not scored — no invented proficiency bars.
01
AI / Machine Learning
- Deployed agentic AI systems at Uber that improved fraud-detection accuracy by 62%+ over manual review, as part of a $30M+ initiative spanning US&C and EMEA.
- One of three model families behind UWFMS, the AI/ML fraud investigation system built at Uber — used alongside decision trees and clustering to cut investigation SLA by 35%.
- Paired with random forests and clustering inside UWFMS, Uber's fraud investigation system, to improve QA accuracy by 20% over the prior manual process.
- Grouped fraud-investigation cases inside UWFMS at Uber; the same technique underpins the Nifty 500 beta-clustering dashboard, which groups stocks into beta regimes via KMeans.
- Benchmarked ARIMA, Prophet, and ML regressors against a naive baseline across the full Nifty 500 universe — and found tree-based models were consistently worse at next-day price prediction.
- Combined with ensemble ML models inside UWFMS / Bliss Scout, the AI fraud-detection platform built at Uber during the Risk & Compliance Specialist II role.
02
Data Science
- Traced a subtle pandas behavior — one NaN silently propagating across an entire rolling-window calculation — while reviewing the 23-indicator backtester's Chaikin Money Flow logic.
- Vectorized the Monte Carlo stock simulator to run 5,000 GBM price paths in roughly 15ms, replacing per-path loops with array-level computation.
- Consolidated three duplicate feature-engineering functions into one while rebuilding the NSE stock-price forecasting benchmark for its multi-model, multi-scale validation.
- Built the 16-gate validation suite behind Apex Quant AI — Information Ratio, Deflated Sharpe, Hansen's SPA, and PBO — that every strategy must clear before reaching portfolio construction.
- Rebuilt statistical and event studies across NSE and cross-market datasets — correlation, seasonality, drawdown — each independently reviewed line by line for fabricated data and dead code.
03
Quantitative Finance
- Built a DCF valuation engine — FCFF, WACC discount, Gordon Growth terminal value — for Nifty 50 constituents, comparing intrinsic value against live market price.
- Ran geometric Brownian motion simulations — 5,000 paths per stock — to generate probabilistic price forecasts with VaR/CVaR tail-risk estimates.
- Backtested 10,902 technical-indicator combinations across roughly 330 NSE stocks since 2009, and separately tested 21,528 strategies for the published SSRN research paper.
- Modeled fraud risk exposure for Uber's Risk & Compliance function, and built the risk engine and validation gates behind the private Apex Quant AI research platform.
- Studying capital markets and trading formally through Masters' Union's PGP in Capital Markets & Trading, on top of ISB's Advanced Management Programme in Business Analytics.
- Authored a 68-page SSRN working paper evaluating technical indicator strategies across Indian and U.S. equities from 2009 to 2026.
04
Product & Systems
- Turned Uber's agentic fraud-detection research into shipped tools with engineering and product partners — scaled across 3 regions while cutting operating costs by ~40%.
- Led the roadmap for a $30M+ agentic AI initiative spanning US&C and EMEA fraud operations at Uber.
- Managed and coached a team of 15+ fraud analysts at Uber while partnering directly with engineering and product — cut average case resolution time by 30%.
- Ran the Bonteous × ISB consulting engagement end-to-end, translating an external client's ambiguous business problem into a structured analytical framework.
05
Tools & Technologies
- Shipped Streamlit dashboards for DCF valuation, Monte Carlo simulation, and Nifty seasonality analysis — with scikit-learn-style regressors underneath the NSE stock-price forecasting benchmark.
- Backs the SQLite storage layer of the NSE market-data monitor — WAL-mode, safe for concurrent reads across 25 tracked data categories.
- Built Tableau dashboards for demand forecasting at Amazon, improving planning accuracy across North America seller operations.
- Paired with Tableau at Amazon to build demand-forecasting dashboards and strengthen financial controls through listing-protection audits.
- Version-controlled every project on this site; the SMC/ICT research repo's commit history documents a real mid-project self-correction after a validation pass caught an overstated result.
06
Domains
- Built UWFMS and led the $30M+ agentic AI initiative at Uber — together spanning fraud investigation, account-takeover prevention, and a 62%+ accuracy gain.
- Subject of a 68-page SSRN research paper and nineteen public and private repositories — from single-stock technical analysis to cross-market strategy backtesting.
- Automated global fraud-detection operations at Uber via agentic AI, and automated financial data collection and alerting via the NSE market-data monitor.
- Served as Risk & Compliance Specialist II at Uber, including a stint as Cyber-Risk SME training 80+ global analysts in AI-driven fraud-detection techniques.
07
Let’s build something that ships.
harshatangirala2@gmail.com- krishna-harsha-tangirala
- Location
- Dubai, UAE
- Website
- krishnaharsha.com