Building high-performance backend architectures, concurrent systems, and scalable software infrastructure.

B.Tech Computer Science undergraduate at MNNIT Allahabad (Class of 2027), passionate about systems programming, concurrent software, and quantitative trading infrastructure.

• Low-latency trading systems
• Concurrent backend architecture
• Quantitative research & market microstructure
• Applied AI & secure agentic architectures

• Languages: C++, Python, MQL5, SQL (Postgres), Java, Bash
• Systems Programming: POSIX, Multithreading, Actor Model, Socket Programming, FastAPI, SQLite
• Applied AI: LangGraph, LLM/SLM Orchestration (Ollama), Context Optimization, Agentic Workflows
• Infrastructure & Security: Docker (Container Sandboxing), Linux, Git, Prometheus, Grafana
• Quantitative & Math: Options Pricing (Black–Scholes, Greeks), Systematic Strategy Backtesting, Volatility-Adaptive Risk Management

An autonomous, multi-threaded, regime-aware options trading system engineered for the Indian derivatives market.

• Decoupled Architecture: Built around a Planner–Executor architecture that isolates analytical workloads from latency-sensitive execution loops using the Actor pattern.
• Memory-Safe Pipeline: Implements a low-overhead concurrent pipeline utilizing auto-trimming collections.deque and thread-safe queues for O(1) ingestion of WebSocket tick data.
• API Protection: Governs broker requests using an asynchronous Token Bucket rate-limiting algorithm within the execution thread to prevent rate-limit bans during market volatility.

A systematic, object-oriented gold (XAUUSD) trading strategy and risk management system built for MetaTrader 5.

• State Recovery: Implements a persistent state recovery system using Terminal Global Variables (GlobalVariableSet) to survive VPS reboots or terminal crashes without desynchronizing open positions.
• Sizing & Risk Engine: Features volatility-adaptive position sizing (ATR-driven), multi-layered drawdown circuit breakers, and programmatic spread checks.
• Backtest Summary: Evaluated across 530 historical trades (99% real tick data, May 2024–May 2026), achieving a 1.37 Profit Factor, 252% net return, and 13.8% maximum drawdown.

An autonomous, event-driven code-remediation agent designed for secure execution environments.

• Sandbox Isolation: Spawns ephemeral, network-disabled Docker containers with strict resource quotas (256MB RAM, 128 PIDs) to safely execute and validate untrusted LLM-generated code.
• Self-Healing Loop: Leverages a LangGraph DAG to recursively run test suites, catch execution errors, and feed stdout/stderr back into the model for iterative patch correction.

• LinkedIn: sri-ram-varun-dittakavi
• Email: sriramvarun636@gmail.com