🛡️
Security & Risk Teams
Building DFIR capabilities, fraud detection pipelines, and compliance workflows that scale with AI-driven orchestration.
AI Security
Enterprise AI Systems Engineer · Gurugram · Available for select engagements
Building Enterprise AI, Platform & Infrastructure Systems.
AI Security · Workflow Orchestration · Quant Finance · Observability · FinOps · Digital Twins — production-grade, fully owned, deployed.
AI Security
Workflow Orchestration
Quant Finance
Observability
FinOps
Digital Twins
12+
Enterprise Systems Built
6
Domains Covered
1×
Live Production Deploy
4
Verified Credentials
🏆 Anthropic Certified 2026
☁️ Oracle GenAI Professional
🔴 Red Hat Certified
Featured Systems
Top differentiators for any recruiter.
If you have 20 seconds on this portfolio, these four systems are the strongest signal of what I build.
🥇
First Place
AegisOps
AI-native fraud detection & DFIR platform for financial voice systems. Multi-agent investigation, deterministic risk engine, audit-chain logging.
AI Security · DFIR · Multi-Agent Systems
🥈
Second Place
Axiom
Enterprise workflow orchestration platform with DAG runtime, event sourcing, distributed state machines, and full observability.
Workflow Orchestration · DAG Runtime · Event Sourcing
🥉
Third Place
QuantEdge AI
Quantitative research platform with backtesting engine, Monte Carlo validation, risk modeling, and statistical analysis pipelines.
Quant Finance · Backtesting · Monte Carlo
🏅
Honorable Mention
IncidentOps AI
AI-powered SRE platform for root cause analysis, anomaly detection, incident correlation, and automated runbook execution.
Observability · SRE · Root Cause Analysis
Who This Is For
Built for operations that run on manual work today.
⚙️
Platform & Infra Teams
Needing workflow orchestration, event-driven DAGs, distributed state machines, and observability platforms.
Platform Engineering📊
Quant & FinTech Teams
Processing large financial datasets. Needing backtesting engines, Monte Carlo validation, and AI-powered analysis.
Quant Finance🔗
Operations-Heavy Businesses
Managing CRMs, ERPs, payment systems. Needing a custom automation and intelligence layer across all tools.
Workflow Automation
Autonomous Systems
AI systems built to automate real business workflows.
Each system replaces a specific manual operation. Code-first, production-grade, fully owned. Click "See It Work" for a live agent simulation.
Revenue Operations · Lead Qualification
Built & Live
LeadPulse
Replaced manual WhatsApp lead follow-up for property sales workflows. Revenue recovery at the automation layer.
- Responds to every inbound lead under 60 seconds — WhatsApp, portals, Meta Ads — without human involvement
- GPT-4o conversational qualification: budget, timeline, intent — hot/warm/cold routing automated
- Hot leads escalated instantly to human closer with full conversation context and CRM record pre-created
- 8-touchpoint 21-day nurture sequence runs autonomously. Site visits booked via Calendar API
<60s
Response
24/7
Coverage
8
Nurture Steps
Zero
Manual Follow-up
Healthcare · Clinic Operations
Built & Live
MediFlow
Eliminated no-shows and manual follow-up for clinic patient workflows.
- T-24hr and T-2hr reminders with one-tap reschedule — no staff involvement
- No-show detection triggers immediate rebooking sequence automatically
- Lab report OCR + structured WhatsApp delivery + critical value alerts to doctor
- Payment dues followed up Day 3, 7, 15 with embedded payment link
-40%
No-Show Rate
0
Manual Calls
Auto
Lab Delivery
Logistics · Last-Mile Delivery
Built & Live
TrackMind
Replaced dispatcher exception calls and manual COD reconciliation.
- Detects failed deliveries within 5 minutes — auto-resolves via WhatsApp interactive buttons
- Customer reschedule flow via button tap — no phone call, no dispatcher
- End-of-shift COD reconciliation: flags mismatches, alerts driver and supervisor automatically
5min
Detection
80%
Auto-Resolved
EOD
COD Reconciled
E-Commerce · Multi-Platform Ops
Built & Live
OrderSync
Replaced manual order management across Amazon, Shopify, and Flipkart.
- Cross-platform order sync, courier booking, anomaly detection — fully automated
- Return processing with fraud signal analysis and pattern detection
- Reimbursement claims for lost/damaged shipments drafted and submitted automatically
30h
Saved/Week
100%
Claims Filed
Auto
Multi-Channel Sync
Field Service · HVAC & AMC
Shipping
FieldOps
Replaced manual dispatch, delayed invoicing, and missed AMC renewals.
- Service request → auto-assigns nearest qualified technician via GPS + skill matching
- Invoice generated and sent the moment technician marks job complete
- AMC renewal sequence launches 60 days before expiry, runs autonomously to payment
0
Missed Renewals
Instant
Invoice
T-60d
Renewal Start
Active Hackathon Builds
Competitive engineering under real constraints.
These are live builds — active hackathon submissions demonstrating production architecture under time pressure.
Active Build
AegisOps
Problem: Financial voice systems are vulnerable to social engineering, verification bypass, and fraudulent wire transfers — with no real-time forensic investigation capability.
⚡ InnovationMulti-agent fraud investigation platform with deterministic risk engine and policy enforcement — detects social-engineering attacks, verification bypass attempts, and fraudulent wire transfers while automatically generating forensic investigations with SHA256 audit-chain logging.
System Architecture · 5 Layers
Layer 1 — Fraud Agent
Multi-agent investigation: real-time call analysis, social engineering detection, verification bypass flagging.
Layer 2 — Deterministic Risk Engine
Policy enforcement with explainable risk scores. No LLM guessing — validated financial risk models only.
Layer 3 — Compliance Agent
Constitutional constraints on agent decisions. Mandatory audit trail. No evidence mutation.
Layer 4 — Hermes DFIR Agent
Automated forensic investigation generation. Incident reports with SHA256 audit-chain logging.
Layer 5 — SOC Dashboard
Real-time event streaming. Live threat monitoring. Explainable security decisions for analysts.
Vestra
Problem: Retail investors have no way to simulate how institutional desks would react to market shocks against their specific portfolio.
Multi-agent LangGraph system. A market-shock agent generates stress scenarios. A risk-validation agent uses deterministic financial models (not LLM guesses) to compute exposure. An execution agent simulates trades with slippage modeling. An explainability agent generates a plain-English risk narrative.
⚡ InnovationDeterministic risk-validation layer before any simulated execution — the LLM can explain but cannot compute risk numbers without passing through validated financial models.
Pit-Wall AI
Problem: F1 race strategy is decided in seconds from high-frequency telemetry. Teams need real-time explainability and scenario simulation — not post-race analysis.
FastAPI WebSocket backend ingests live telemetry. Counterfactual simulation engine generates "pit now" vs "stay out" scenarios with probability-weighted outcomes. LangGraph manages strategy decision state across pit windows. Tire degradation models and safety-net data sources integrated.
⚡ InnovationAI race engineer that generates real-time pit stop recommendations, tire degradation analysis, and scenario analysis — fully deployed in production.
Enterprise Systems Portfolio
Platform and infrastructure systems across six domains.
Beyond automation — these are full enterprise-grade systems spanning AI security, workflow orchestration, quantitative finance, observability, FinOps, and supply chain simulation.
AI Security · DFIR
Active Build
AegisOps
AI-native fraud detection and DFIR platform for autonomous financial voice systems.
- Multi-agent fraud investigation with Fraud, Compliance & Hermes DFIR agents
- Deterministic risk engine with policy enforcement and explainable decisions
- SHA256 audit-chain logging and automated DFIR report generation
- Real-time SOC dashboard with live event streaming
Workflow Orchestration
In Progress
Axiom
Enterprise workflow orchestration platform with DAG runtime and event sourcing.
- DAG execution engine with distributed state machines and dependency resolution
- Event sourcing architecture with full audit trail and replay capability
- Workflow orchestration with conditional branching and error compensation
- Distributed tracing and observability built-in
Quant Finance
In Progress
QuantEdge AI
Quantitative research platform with AI-powered backtesting and Monte Carlo validation.
- Backtesting engine with slippage modeling and execution simulation
- Monte Carlo simulation for risk modeling and statistical validation
- Time-series forecasting with regime detection
- LLM-powered market narrative generation and explainability
Observability · SRE
In Progress
IncidentOps AI
AI-powered observability and SRE platform for root cause analysis and automated remediation.
- Anomaly detection with multi-signal correlation across metrics, logs, and traces
- Root cause analysis using causal AI and dependency graph traversal
- Automated runbook execution with guardrailed remediation
- OpenTelemetry integration with Prometheus, Grafana, and Loki
FinOps · Cost Optimization
In Progress
CloudCostIQ
Cloud cost intelligence platform with AI-driven optimization recommendations and forecasting.
- Multi-cloud cost ingestion with anomaly detection and budget alerting
- AI-powered optimization recommendations with ROI quantification
- Cost forecasting with scenario modeling and budget trend analysis
- Showback and chargeback reporting for engineering teams
Simulation · Operations Research
In Progress
Supply Chain Digital Twin
AI-powered supply chain simulation platform with disruption modeling and operations research.
- Digital twin simulation of end-to-end supply chain with real-time state updates
- Disruption scenario modeling: demand shocks, supplier failures, logistics delays
- Operations research optimization for inventory and routing decisions
- AI Copilot for supply chain analysts with natural language querying
What You Get
Every engagement includes full ownership and handover.
No black-box deliveries. No platform lock-in. Every system I build is yours — source code, documentation, deployment scripts, and monitoring. You can run it, modify it, or hand it to any developer after I'm done.
✓
Source Code Ownership
Full repository access. All code is yours on delivery.
✓
Full Documentation
Architecture docs, API references, deployment guides.
✓
Deployment Support
Docker, CI/CD, cloud setup — running in production.
✓
All API Integrations
WhatsApp, CRM, payment, calendar — wired and tested.
✓
Monitoring Setup
Grafana dashboards, Prometheus metrics, error tracking.
✓
Complete Handover
Walk-through session. You understand every part of what was built.
About
Enterprise AI, Platform & Infrastructure Systems across six domains.
From AI-native fraud detection platforms to enterprise workflow orchestration, quantitative research systems, and supply chain digital twins — I build production-grade systems that think, decide, and execute at scale. Every deployment is code-first, fully owned, and architected for real enterprise requirements.
I'm Mayank Sharma, currently based in Gurugram and working as a Quantitative Researcher at Cite Sert, where I build LLM-powered financial NLP pipelines for NSE market intelligence. I hold a verified Anthropic certification in agentic AI systems alongside Oracle and Red Hat credentials.
🛡️
AI security: multi-agent DFIR, deterministic risk engines, audit-chain logging, compliance policy enforcement
⚡
Platform engineering: DAG runtimes, event sourcing, distributed state machines, Kafka event streams
📊
Quant finance: Monte Carlo simulation, backtesting, risk modeling, time-series analysis, LLM narration
🔍
Observability: OpenTelemetry, Prometheus, Grafana, Loki, root cause analysis, AI-powered SRE
Engineering Systems
ML infrastructure & deployed backends.
Production APIs, financial NLP systems, and full-stack ML deployments demonstrating technical depth beyond agent work.
📡
Quantitative Research · Financial NLP · NSE
MarketPulse
LLM pipeline that ingests financial news and maps each article to materially relevant NSE-listed companies. GPT-4o contextual extraction + deterministic validation against the NSE universe. Handles ticker disambiguation, abbreviation resolution, false-positive suppression.
⌥ View on GitHub →⚙️
ML API · FinTech · Production
InsurPredict
Regression model on 15,000+ records. R² 0.86, deployed as a REST API at 500+ req/min. Pydantic validation at the API layer cut invalid submissions by 40%.
⌥ View on GitHub →🛡️
ML · Financial Security
FraudGuard-AI
ML pipeline for financial fraud classification on imbalanced datasets. Ensemble methods, precision-recall benchmarking, structured feature engineering.
⌥ View on GitHub →🧬
ML · Healthcare · Classification
DiaPredict-AI
Clinical ML model predicting diabetes risk. Full pipeline: preprocessing, feature selection, cross-validation, hyperparameter tuning, interpretability analysis.
⌥ View on GitHub →🎬
Recommender System · NLP
Movie Recommendation Engine
Content-based recommender on the TMDB 5000 dataset. TF-IDF vectorization + cosine similarity over film metadata. 5,000+ titles with a personalized query interface.
⌥ View on GitHub →🖥️
Full-Stack ML · Deployment
ML Prediction App
Decoupled FastAPI inference backend + Streamlit dashboard. 10,000+ rows. Fully Dockerized. Clean API contract between the ML and UI layers.
⌥ View on GitHub →
Experience
Applied AI in production environments.
Quantitative Researcher
Cite Sert · Remote
MAY 2026 – NOV 2026
Agentic AI & Financial NLP
- Architected an LLM-powered financial news entity-mapping pipeline using contextual extraction + deterministic validation against the NSE universe — producing structured symbol-mapped datasets for downstream alpha generation
- Engineered hybrid NER + symbol resolution handling ticker disambiguation, abbreviation expansion, and false-positive suppression — mapping entities to exact NSE symbols and company names
- Benchmarked structured prompt architectures for precision, recall, and consistency across 100+ financial news classifications
Finance Research Analyst Intern
Edify Equity · Remote
APR 2026 – MAY 2026
Equity Research
- Produced weekly intelligence briefs synthesizing equity research, macro indicators, and financial news into structured summaries for senior leadership
- Maintained systematic market records and financial process documentation — establishing the domain knowledge that informs all financial AI work
Process
Workflow description to deployed system.
01
🔍
Map the Workflow
You describe what your team does manually. I map every step, decision, and failure point before proposing anything.
02
📐
Design the System
State machine design, decision logic, tool calls, escalation rules — documented and approved before a line of code is written.
03
⚙️
Build Production-Grade
LangGraph pipelines, FastAPI backends, Redis state, retry logic, dead-letter queues. Built to handle real edge cases.
04
📊
Deploy & Handover
Docker, CI/CD, monitoring dashboards, walk-through session. You own everything. Accountable to the agreed outcome metric.
Tech Stack
Production tools across the full enterprise stack.
Core
Proficient
Familiar
// AI & Agentic Systems
LangGraphUsed inAegisOps · LeadPulse · Vestra · Pit-Wall AI
LangChainUsed inAgent tracing & pipeline development
Multi-Agent SystemsUsed inAegisOps · Vestra · QuantEdge
Agent OrchestrationUsed inFraud, Compliance & Hermes agents
RAG PipelinesUsed inFinancial intelligence systems
Prompt EngineeringUsed inEvery agent & NLP pipeline
Vector SearchUsed inRAG & semantic retrieval pipelines
Tool CallingUsed inAegisOps · LeadPulse · Axiom
AI CopilotsUsed inCloudCostIQ · Supply Chain Twin
LLM EvaluationUsed inMarketPulse · prompt benchmarking
GPT-4o · GeminiUsed inLeadPulse · OrderSync · MarketPulse
Claude API · MCPUsed inMCP context server projects
// Backend & APIs
PythonUsed inAll systems & pipelines
FastAPIUsed inEvery backend · InsurPredict · ML App
REST APIsUsed inAll production systems
PydanticUsed inMarketPulse · InsurPredict
SQLAlchemy · AlembicUsed inDatabase ORM & migrations
JWT AuthenticationUsed inLeadPulse · AegisOps · Axiom
RBACUsed inEnterprise systems security layer
Background WorkersUsed inCelery · async job queues
// Data & Databases
PostgreSQLUsed inAll production systems
RedisUsed inLeadPulse · AegisOps · TrackMind
pgvectorUsed inVector search & RAG pipelines
Data ModelingUsed inAll system design phases
Database OptimizationUsed inProduction query tuning
Pandas · NumPyUsed inAll data pipelines
// Distributed Systems
KafkaUsed inAxiom event streaming
Event-Driven ArchitectureUsed inAxiom · AegisOps
Workflow OrchestrationUsed inAxiom DAG runtime
DAG ExecutionUsed inAxiom workflow engine
Event SourcingUsed inAxiom audit trail architecture
State MachinesUsed inLangGraph nodes · Axiom
Distributed TracingUsed inIncidentOps AI · OpenTelemetry
// ML & Data Science
Scikit-LearnUsed inInsurPredict · FraudGuard · DiaPredict
NLP · NERUsed inMarketPulse · Movie Engine
Statistical ValidationUsed inQuantEdge AI backtesting
PyTorchUsed inML experimentation
// Cloud & DevOps
DockerUsed inInsurPredict · ML App · all deployments
KubernetesUsed inContainer orchestration
GitHub Actions CI/CDUsed inEvery deployment pipeline
AWS EC2 · ECSUsed inProduction hosting
TerraformUsed inInfrastructure as code
// Observability
Prometheus · GrafanaUsed inMonitoring & alerting
OpenTelemetryUsed inIncidentOps AI distributed tracing
LokiUsed inLog aggregation · IncidentOps AI
// Quant & Analytics
Monte Carlo SimulationUsed inQuantEdge AI risk validation
Risk ModelingUsed inQuantEdge AI · Vestra
Time-Series AnalysisUsed inQuantEdge AI · Pit-Wall AI
ForecastingUsed inCloudCostIQ · QuantEdge
// Specialized Domains
DFIRUsed inAegisOps forensic investigation
Fraud DetectionUsed inAegisOps · FraudGuard-AI
FinOpsUsed inCloudCostIQ cost platform
Operations ResearchUsed inSupply Chain Digital Twin
WhatsApp Business APIUsed inLeadPulse · MediFlow · TrackMind · FieldOps
Credentials
Verified certifications.
🏆
Anthropic
Building with the Claude API
Agentic AI, MCP, Prompt Engineering
2026 · Verified
☁️
Oracle
Cloud Infrastructure Generative AI Professional
LLMs, RAG, GenAI Agents
2025 · Verified
☁️
Oracle
Cloud Infrastructure Data Science Professional
ML Workflows, Model Deployment
2025 · Verified
🔴
Red Hat
System Administration I & II
RH124 & RH134
2025 · Verified
Contact
Need an AI system for your workflow?
Let's map your automation in 15 minutes. Describe what your team does manually today — I'll tell you what a custom agent can take over, what stack I'd use, and how long it takes.