ATHARVA TRIVEDI
🇮🇳 Bangalore, India
- TeamGlobalLogic → Ericsson
- ChampionshipAI / ML Engineering
- ChassisB.Tech CSE, VNIT Nagpur
- Debut2021 · Full-time 2025
- Best Lap−31% signalling (5G advisor)
FORMATION LAP…
🏁 ROUND 1 · BANGALORE GP · 2026 SEASON
AI/ML SOFTWARE ENGINEER
I build LLM agents, RAG pipelines, and full-stack systems — currently racing for Ericsson's 5G RAN program. Bit of a cat about it.
🇮🇳 Bangalore, India
AI/ML-focused software engineer who likes the corners where backend engineering and agentic AI meet.
I've shipped production LLM agents and RAG pipelines, launched a full travel app solo to the App Store and Play Store, and I'm now applying that to Ericsson's 5G RAN program — driving Layer 2 downlink test automation with agentic AI tools and owning the Physical AI side of a robotics Centre of Excellence. I ramp into new technical domains fast and turn that into measurable impact.
Off the clock I'm the kind of engineer who'd happily name a variable
lewis and treats a green CI pipeline like a Hamilton win at
Silverstone. Curious, a little competitive, and always looking for the next apex.
GlobalLogic — Ericsson 5G RAN Program
Aug 2025 — Present
Montty
May 2026 — Present
AIonOS
Jul 2024 — Jan 2025
AlphaFi
Apr 2024 — Jun 2024
Click2Cloud, Nagpur
Dec 2023 — Jan 2024
ClipFusion
May 2023 — Jul 2023
Click2Cloud, Nagpur
Dec 2022 — Jan 2023
Production travel platform · sole engineer
Shipped a full travel app end-to-end as the sole engineer — design through App Store and Play Store launch — with real-time chat (Stream), voice/video calls (Agora), live location (Mapbox), a Groq-powered itinerary planner, and visa-application workflows on a cloud-native Firebase/Supabase backend.
C11 + fixed-point ML · on-device
An LTE RRC base-station state machine in C11 with a self-contained ML "release advisor". Cuts control-plane signalling by 31% at matched connected-time versus a static timer — with zero illegal state transitions. My 5G RAN day-job, made concrete.
Agent-based modeling · ISCON 2025
Combined evolutionary game theory with Stackelberg optimization in an agent-based simulation — ~25% lower tumor burden and ~28% less treatment intensity. Presented the findings at ISCON 2025.
🏷️ PRIVATE SALE · ENQUIRIES OPEN
I designed, built, and shipped a product end to end — solo. It's real, it works, and it's ready for an owner who can take it further than I can on my own. If you're after a working app with room to grow, the door's open.
Press the button. I promise nothing.
Some days the car catches fire. Tap to open the incident report.
tap to reveal ▸
The time I set up an infinite pull loop on Groq's API and burned through every token in my quota before my morning coffee — just staring at the cursor blinking while tokens flew. Turned out I'd forgotten a break condition in my streaming parser. My Groq dashboard went to $0.00 in 47 seconds. Learned to always test with max_tokens=5 first.
Beat a real F1 start. Wait for lights out, then tap. Jump the start and it's a penalty.
Got a seat for me, a project, or an offer on the app? Open the channel.