Technical Lead, Gurugram, India
Open to opportunitiesI build distributed systems and platform infrastructure that hold up at scale.
8+ years across startups and enterprises. Currently leading a 12+ engineer team at SYMX.AI. Previously at Thales and LambdaTest.
I specialize in IoT observability, Go microservices, and cost-aware platform engineering. Based in Gurugram, working with global teams.
8+ years of engineering across distributed systems, platform infrastructure, and security.
Technical Lead
Jan 2025 - Present
Leading a 12+ engineer team building X-Radar, a real-time IoT observability platform reading data from 1,500+ Symbots on mining trucks, and Control Room for live fleet tracking.
Senior Software Engineer
Jun 2021 - Jan 2025
Architected 5 Go microservices for CipherTrust Secrets Management on Kubernetes; integrated CipherTrust with Akeyless. Won the global Thales CPL Award.
Software Engineer
Nov 2020 - May 2021
Built Node.js backend services for RainierAEV, an ML-driven autonomous testing platform for enterprise ERP systems including Workday, SAP, and Oracle.
Member of Technical Staff
Nov 2019 - Oct 2020
One of the first 30 employees. Built integrations with 15+ Selenium, Playwright, and Cypress frameworks; published a Java Maven plugin to Maven Central.
Associate Software Engineer
Jan 2018 - Oct 2019
Java feature development and bug fixes for Bell Canada's telecom systems.
Platforms, tools and open source, built to solve real problems.
Independent research project that reads the bills of AI, cloud, and API providers and attributes every dollar of spend to the team, feature, or customer behind it. No SDK, no code changes.
Open-source VS Code extension that scans codebases for exposed secrets (API keys, passwords, tokens) and raises real-time alerts with customisable detection rules.
Open-source Chromium extension published under the ThalesGroup GitHub org. Automates SSO configuration between CipherTrust Manager and Akeyless, cutting enterprise setup time from 30 minutes to under 30 seconds.
Senior Technical Lead with 8+ years building distributed systems and platform infrastructure. Currently leading a 12+ engineer team at SYMX.AI.
I specialize in engineering distributed systems, platform infrastructure, and developer tooling. At SYMX.AI I built X-Radar from scratch: a real-time IoT observability platform reading data from 1,500+ Symbots on mining trucks, cutting customer escalations by 89 percent.
At Thales I architected Go microservices for CipherTrust Secrets Management and won the global CPL Award. I also build open-source tools like SecretStack and the CipherTrust SSO plugin. Based in Gurugram, India, working with global teams.
B.Tech, Computer Science
GL Bajaj Group of Institutions, AKTU (2014–2018) · 85.2% · College Topper · Top 30 in UP
Engineering deep-dives, architectural patterns, and notes from building systems in production.
Deployment strategies compared: rolling updates, blue-green, and canary. Traffic shifting, rollback, and readiness gates.
Inside X-Radar: the real architecture of a Streamlit observability platform processing data from 1,500+ mining truck devices across Athena, DynamoDB, and PostgreSQL.
How to build webhook delivery that works: at-least-once semantics, retry with backoff, signature verification, and ordering guarantees.
Model costs, infrastructure overhead, and margin math. How to calculate whether your AI feature is actually profitable.
InfluxDB, TimescaleDB, or plain S3 for time-series data? A practical comparison of write throughput, query patterns, retention, and cost at scale, with a decision framework.
Technical leadership, delegation, architecture ownership, and stakeholder management. What actually changes when you move from senior engineer to tech lead.
Evaluation rubrics, common mistakes, and how to structure answers. What separates a strong senior system design interview from an average one.
Package layout, dependency direction, interface design, and the monorepo question, drawn from running multiple Go microservices in production.
JSON logging, correlation IDs, log levels, cardinality control, and cost management for production microservice logging.
Secret injection patterns, rotation without restarts, and audit trails: production secrets management compared.
How to handle transactions across microservices: saga pattern HLD, choreography vs orchestration, compensation logic, and isolation challenges.
Generic AI assistants write plausible code that ignores your conventions. Repo-specific context files and mandatory human review are what make AI output shippable.
Measuring toil, building self-service, and paving roads without becoming the team everyone waits on.
Low-level design of a real-time alerting pipeline: stream processing, windowing, alert deduplication, and latency budgets from sensor to notification.
Type systems, error handling, deployment, and performance characteristics compared from experience running both in production.
Event loop blocking, connection pool sizing, structured concurrency, and debugging techniques for Python asyncio services that actually ship.
Platform HLD, golden paths, escape hatches, and the line between helpful abstraction and harmful magic.
HLD of an OpenTelemetry pipeline: instrumentation, collectors, backends. Traces vs metrics vs logs, sampling strategies, context propagation, and the real tradeoffs nobody warns you about.
A complete HLD for multi-tenant SaaS: isolation models, noisy neighbor handling, and what to do when one tenant is 100x larger than the rest.
Cluster per environment vs per team vs per region: multi-cluster patterns, federation, and GitOps approaches.
Model Context Protocol servers in production: HLD, auth, sandboxing tool execution, input validation, failure modes, scaling, and observability. An engineer's guide.
Token metering, per-request tagging, cost allocation models, and budget guardrails. How to know exactly which feature is burning your OpenAI budget.
Running inference and microservices on shared Kubernetes clusters is becoming normal. GPU scheduling, tenant isolation, and idle-cost control decide whether it works.
A practical guide to cutting Kubernetes spend: right-sizing workloads, autoscaling that responds to real demand, spot instances without the pain, and placement strategies that bin-pack efficiently.
Throughput, ordering, retention, operational burden, and cost: a practical framework for choosing between Kinesis and Kafka.
HPA vs KEDA vs custom metrics: scaling signals, cooldown behavior, and when each approach makes sense.
Mining sites lose connectivity. Your IoT architecture needs offline buffering, store-and-forward, conflict resolution, and the right MQTT QoS levels to survive it.
How to design APIs that safely handle retries: idempotency keys, exactly-once myths, and implementation patterns that work.
Latency benchmarks, schema evolution, tooling tradeoffs, and a practical framework for choosing between gRPC and REST for service-to-service communication.
Signal handling, draining in-flight requests, shutdown ordering, and Kubernetes termination lifecycle for Go services.
Worker pools, fan-in/fan-out, context cancellation, errgroup, and pipeline patterns you will actually use in production Go services.
How pod networking actually works on EKS: VPC CNI modes, security groups for pods, network policies, and DNS.
When distributed locks are necessary, how Redlock and Zookeeper work, fencing tokens, and the alternatives you should consider first.
High-level design of an event-driven microservices platform in Go: event bus architecture, schema evolution, dead letter queues, and ordering guarantees.
A production-grade HLD for AI agent platforms: orchestrator, tool router, per-session sandboxes, memory stores, and observability. Isolation tiers, secrets, egress, and cost controls included.
Doc structure, tradeoff analysis, alternatives considered, and decision records. How to write technical design documents that actually drive decisions.
A practical decision framework for database scaling: sharding strategies, shard key selection, read replica lag, and when each approach makes sense.
Lambda vs kappa architecture, exactly-once semantics, late data handling, and backfill strategies. A practical guide to choosing between batch and streaming pipelines.
Pool sizing math, timeout tuning, exhaustion handling, and when to reach for PgBouncer in production backend services.
What AI gets wrong: edge cases, error paths, hidden assumptions. A practical review checklist for AI-assisted pull requests.
AWS bills show services and accounts, not features or teams. Here is why the bill structure makes attribution hard, and how to build a pipeline that fixes it.
How to prevent cascading failures: circuit breaker states, retry budgets, timeout design, bulkheads, and hedging.
Pipeline HLD for microservices: trunk-based development, progressive delivery, and DORA metrics that matter.
Practical cache invalidation strategies for distributed systems: write-through, TTL, event-driven invalidation, and stampede protection.
Job queue HLD, at-least-once delivery, dead letter handling, scheduling, and priority queues for reliable background processing.
What causes Lambda cold starts, how to mitigate them, and when serverless is the wrong choice.
URL vs header versioning, deprecation policies, additive changes, and contract testing for evolving APIs without breaking consumers.
High-level design of an API gateway: request routing, rate limiting algorithms, auth termination, and request transformation at scale.
Agents that execute code are remote execution services with a natural language interface. Isolation, identity, and audit trails are infrastructure problems, not model problems.
Coding agents burn tokens the way servers burn CPU, but most teams still cannot say where the tokens go. Token metering, budget guardrails, and per-team attribution are how you fix that.
Open to high-impact technical leadership roles. I usually reply within a day.
arpitstack@gmail.com