Software engineer working on AI inference and model optimization, with a systems, backend, and performance-engineering foundation underneath it. I care about how things work at the level below the framework — memory layout, concurrency, and where the cycles actually go.

Day to day right now: inference/optimization work (current focus), plus Java/Spring Boot backend work on the side to stay sharp on distributed systems fundamentals.

C++ (11/14/17/20) · Data Structures & Algorithms · Multithreading & Concurrency · Linux & Systems Programming · Memory Management · Networking · Performance Optimization · Java · Spring Boot · Kafka · Redis · PostgreSQL · Python · Distributed Systems

• OS-COMPONENTS-CPP — a collaborative group project building core OS components in C++: a memory allocator (wrapped in an Allocator class), a CPU process scheduler (FCFS/SJF/Round Robin/priority with aging), a cooperative user-space thread library (context switching, join/exit, its own round-robin scheduler), and IPC (pipes, message queues) so far, with virtual memory, a file system, and a shell still to come. Forked from PoojaGoel-IIT/OS-COMPONENTS-CPP.
• url-shortener — a URL-shortening service on Spring Boot 3.3 / Java 21 with PostgreSQL, a proper layered package structure, and a JUnit/MockMvc/H2 test suite.
• differential-Privacy-using-coreset — coreset-based sampling for privacy-preserving, compute-efficient clustering; measures approximation error vs. random sampling.
• Core-Machine-leaning — active learning / submodularity for informative subset selection, benchmarked across KNN, Logistic Regression, SVM, and Random Forest.
• Brain-tumor-detection- — CNN-based brain tumor classification from MRI scans (Keras/TensorFlow).

System_design is where most of my C++ systems practice lives:

• Multithreading primitives built from scratch: a working thread pool, ring buffers (including a multi-producer/multi-consumer variant), producer-consumer, atomic flags.
• Hand-rolled STL pieces (unique_ptr, shared_ptr, a dynamic array) to understand what the standard library is doing under the hood.
• Two full low-level-design exercises: a Parking Lot system and an Elevator System, each split into single-responsibility classes.
• C++17 features (optional, variant, any) and classic design patterns (Singleton, Observer, Factory, Abstract Factory).

• url-shortener — Spring Boot + PostgreSQL, tested end to end.
• java — an Order/Inventory Management backend built iteratively (Spring Boot, JPA/Hibernate, JWT auth, Redis caching, optimistic locking with @Version), plus a modular Parking Lot LLD exercise.
• Comfortable with Kafka and Redis for the messaging/caching side of distributed systems, applied here at practice scale.

Competitive-Programming — solutions across LeetCode, Codeforces, and CodeChef, plus reusable DSA templates (Union-Find, monotonic-stack next-greater, segment tree).

• LeetCode rating: 1850+
• HackerRank: 5-star (C++)
• GATE CS: 99th percentile

AI inference optimization techniques, and going deeper on distributed systems design (the Kafka/Redis side of things beyond toy examples).

• LinkedIn: shivam-mishra-3242361aa
• Email: mishrashivam@alumni.iitgn.ac.in