--- title: "Shivanshu Sharma - Research Engineer 1 - C++ Platform and Data Systems in Bengaluru | Tal" description: "Meet Shivanshu Sharma, Software Engineer @ LG Soft India | C++17/20 · Low-Latency Systems · Lock-Free Concurrency · Linux/UNIX · IPC & Distributed Systems | Building high-throughput time-series and data-serving platforms based in Bengaluru. AI stack includes anthropic apis, claude code, and langcha..." image: "https://tal.club/meet/shivanshu-sharma-1/share-image" --- # Shivanshu Sharma on Tal ## Headline Software Engineer @ LG Soft India \| C\+\+17/20 · Low\-Latency Systems · Lock\-Free Concurrency · Linux/UNIX · IPC & Distributed Systems \| Building high\-throughput time\-series and data\-serving platforms ## About I build C\+\+ systems where latency, correctness, and uptime are all non\-negotiable at once\. Two years into production engineering at LG Soft India \(LG Electronics R&D\), most of my work sits close to the metal: core Radio Interface Layer components in C\+\+17, UNIX domain sockets for low\-latency IPC between telephony and connectivity processes, message framing, and connection lifecycle management inside a stack that ships in real vehicles\. Alongside that I carry Level 3 production support — the correctness bugs, race conditions, memory defects, and performance regressions land on me and stay there until they're genuinely understood, not just suppressed\. That has taught me more about how systems actually behave than any amount of reading\. I'm also the sole owner of a PostgreSQL\-backed data\-serving platform used daily by 200\+ engineers, where I designed the schema, transactional workflows, query paths, migrations, and a zero\-downtime deployment pipeline on GitHub Actions\. Owning something end\-to\-end — including the 2am failure modes — changes how you design it\. WHAT I WORK ON OUTSIDE OF WORK • A zero\-copy SPSC ring\-buffer ingestion pipeline in C\+\+17 sustaining 1M\+ messages/sec at deterministic 200–800ns per\-message latency, with cache\-efficient contiguous memory layouts and automated latency\-percentile regression detection\. • A live C\+\+20 compute engine consuming real\-time time\-series data from multiple independent exchange endpoints over concurrent WebSocket feeds, hitting \~10µs tick\-to\-signal on the critical path, with deterministic state machines and crash recovery so it runs unattended\. • An LLM\-based log analysis platform \(Python/asyncio\) routing 20,000\+ events through cost\-aware pipelines for automated root\-cause investigation, validated against verified production defects\. HOW I WORK I use AI\-assisted tooling seriously — Claude Code and the Anthropic/OpenAI APIs are part of my daily development loop, gated by automated tests and peer review rather than trusted blindly\. I care about mechanical sympathy: knowing what the cache, the allocator, the scheduler, and the kernel are actually doing underneath the abstraction\. Currently going deeper on distributed storage — columnar formats, replication, partitioning, and consensus\. Open to conversations about low\-latency systems, distributed infrastructure, and trading technology roles\. 📧 shivanshu\.0710@gmail\.com ## Location Bengaluru ## AI stack anthropic apis claude code langchain langgraph langsmith notebooklm ## Work experience Research Engineer 1 \- C\+\+ Platform and Data Systems at LG Soft India, Mar 2026 \- Present · 7 mos, Bengaluru, Designed and implemented a fully local, multi‑agent LLM pipeline that autonomously performs root‑cause analysis on Anritsu RTD test failures using QXDM OTA logs, eliminating manual triage time\. Research Engineer at LG Soft India, Apr 2024 \- Mar 2026 · 2 yrs, Bengaluru, Core C\+\+ engineering on automotive connectivity systems, plus ownership of an internal data\-serving platform\. • Build core Radio Interface Layer \(RIL\) components in C\+\+17 using UNIX domain sockets for low\-latency IPC between telephony and connectivity processes — message framing, connection lifecycle, and reliable communication across process boundaries in a shipped production stack\. • Reduced critical\-path latency by 20\.5% in production C\+\+ services via lock\-free queues, cache\-line\-aware data layout, and hot\-path refactoring, validated with reproducible benchmark harnesses\. • Sole owner of a PostgreSQL\-backed data\-serving platform processing structured engineering datasets for 200\+ engineers daily — schema design, transactional workflows, query optimization, migrations, and zero\-downtime deployment via GitHub Actions\. • Own Level 3 production support for business\-critical platforms: root\-cause and resolve correctness bugs, race conditions, memory defects, and performance regressions directly from production traces\. • Built a high\-throughput asynchronous event\-processing pipeline in Python \(asyncio\) handling 20,000\+ telecom log events per run with fault tolerance, retry semantics, and structured observability\. • Integrated AI\-assisted development tooling \(Claude Code, Anthropic/OpenAI APIs\) into the team SDLC for code generation, large\-scale refactoring, and debugging, with automated testing and peer review as correctness gates\. Stack: C\+\+17/20 · Linux/POSIX · UNIX domain sockets · IPC · multithreading · lock\-free data structures · PostgreSQL · Python \(asyncio\) · Docker · GitHub Actions · AWS EC2 Contract Software Engineer at LG Soft India, Apr 2024 \- Feb 2026 · 1 yr 11 mos, Converted to permanent ## Education Bachelor of Engineering \- BE, Computer Science and Engineering, Chitkara University High School, Science, Him Academy Public School ## Public links - github: https://github\.com/shivanshu1177 [View the original profile](https://tal.club/meet/shivanshu-sharma-1) ```json { "@context": "https://schema.org", "@graph": [ { "@type": "Person", "@id": "https://tal.club/meet/shivanshu-sharma-1#person", "name": "Shivanshu Sharma", "url": "https://tal.club/meet/shivanshu-sharma-1", "image": "https://tal.club/meet/shivanshu-sharma-1/profile-photo.jpg", "description": "Meet Shivanshu Sharma, Software Engineer @ LG Soft India | C++17/20 · Low-Latency Systems · Lock-Free Concurrency · Linux/UNIX · IPC & Distributed Systems | Building high-throughput time-series and data-serving platforms based in Bengaluru. AI stack includes anthropic apis, claude code, and langcha...", "jobTitle": "Research Engineer 1 - C++ Platform and Data Systems", "worksFor": { "@type": "Organization", "name": "LG Soft India" }, "hasOccupation": { "@type": "Occupation", "name": "Research Engineer 1 - C++ Platform and Data Systems" }, "homeLocation": { "@type": "Place", "name": "Bengaluru" }, "alumniOf": [ { "@type": "EducationalOrganization", "name": "Chitkara University" }, { "@type": "EducationalOrganization", "name": "Him Academy Public School" } ], "knowsAbout": [ "anthropic apis", "claude code", "langchain", "langgraph", "langsmith", "notebooklm" ], "sameAs": [ "https://github.com/shivanshu1177" ], "mainEntityOfPage": { "@id": "https://tal.club/meet/shivanshu-sharma-1#webpage" } }, { "@type": "ProfilePage", "@id": "https://tal.club/meet/shivanshu-sharma-1#webpage", "url": "https://tal.club/meet/shivanshu-sharma-1", "name": "Shivanshu Sharma on Tal", "description": "Meet Shivanshu Sharma, Software Engineer @ LG Soft India | C++17/20 · Low-Latency Systems · Lock-Free Concurrency · Linux/UNIX · IPC & Distributed Systems | Building high-throughput time-series and data-serving platforms based in Bengaluru. AI stack includes anthropic apis, claude code, and langcha...", "dateModified": "2026-09-15T01:49:52.125Z", "isPartOf": { "@id": "https://tal.club/#website" }, "mainEntity": { "@id": "https://tal.club/meet/shivanshu-sharma-1#person" }, "primaryImageOfPage": { "@type": "ImageObject", "url": "https://tal.club/meet/shivanshu-sharma-1/profile-photo.jpg", "contentUrl": "https://tal.club/meet/shivanshu-sharma-1/profile-photo.jpg", "caption": "Shivanshu Sharma profile photo on Tal" }, "inLanguage": "en-IN" } ] } ```