Haarvish Chevula
Founder, Ferronyx · Researcher, SORA
· Software Engineer, Sanas · Bengaluru, India
About
Haarvish Chevula builds runtime intelligence for autonomous
systems — the observability, reasoning, and recovery layers
that let machines run themselves when the operator can't
answer in time. On servers. On robots. In orbit.
Ferronyx — The intelligence layer for robotics
Ferronyx is a ROS2 observability platform with AI-assisted
root cause analysis and fleet-wide diagnostics. It reduces
robot MTTR from hours to minutes through real-time topic
graphs, autonomous correlation across nodes and sensor
streams, and remote triage that doesn't require flying an
engineer to the field.
Visit Ferronyx
SORA — Spacecraft Observability and Runtime Autonomy
SORA is an AI-native, domain-specialized health management
and lifecycle autonomy framework for space systems.
Co-researched with Sahasra Boyapati (Embry-Riddle
Aeronautical University, NASA SBIR / RIDDANCE).
Status: active R&D. Validation against NASA SMAP and MSL
labeled-anomaly telemetry benchmarks in progress. Full
architecture and specifics are restricted pending
publication. Brief available to qualified partners under NDA
at haarvish@gmail.com.
Sanas — Software Engineer (2025 — present)
Leads platform-wide performance engineering for a real-time
voice AI system under production load. Instrumented 100% of
microservices with New Relic, establishing full-stack METL
(Metrics, Events, Logs, Traces) observability. Runs
benchmarking and infrastructure profiling against peak load.
Ships end-user-impacting fixes across application code,
inter-service communication, cloud infrastructure, and
autoscaling.
Temperstack — Founding Engineer (2023 — 2025)
Built and scaled end-to-end AI infrastructure and reliability
tooling. Designed and shipped an AI-powered SRE agent on
LangGraph for smart runbook execution and autonomous
incident resolution — reduced customer MTTR ≈45%. Led alert
correlation and auto-resolution engines, a self-healing
daemon framework for infrastructure reliability, and
integrations across AWS, GCP, Azure and APMs (New Relic,
Datadog, AppDynamics, Dynatrace, Splunk).
Earlier work
-
AMS India — Computer Vision Developer
(2022–2023)
-
Mirasys — Student Intern, Deep Vision
(2021–2023)
-
Youth India Foundation — Management
Intern (2021)
Education
Jain (Deemed-to-be University) — B.Tech,
Mechatronics, Robotics & Automation Engineering (2019–2023)
Thesis
Observability and runtime autonomy aren't three different
problems at three different altitudes — they're the same
problem in progressively harsher conditions. At the server
tier the stakes are latency and revenue. At the robot tier
they're uptime and physical safety. At the spacecraft tier
they're mission loss and an orbital commons that cannot be
cleaned up after the fact. The engineering is the same
shape at every altitude: ingest heterogeneous telemetry,
detect the anomaly, isolate the root cause, forecast
degradation, act autonomously when a human can't respond in
time.