Solution architect and Go engineer building edge infrastructure for camera networks and 5G platforms.

I design and build 5G/MEC video intelligence platforms at EPAM Systems: RTSP/ONVIF camera pipelines, real-time edge inference, and secure device management across field deployments.

Greater Boston. The repos here are the same problems worked in the open.

13+ yrs infra 20+ tier-one telecom clients ~50 devices deployed ≤500ms e2e latency

↓ resume linkedin.com/in/ritwikjain github.com/0x524a credly

about

Thirteen years ago I was debugging S1AP failures deep inside LTE core infrastructure. Today I'm architecting GPU inference pipelines at the edge of that same network. Most engineers specialize on one side of that line (protocol internals or AI/ML) because the two rarely sit in one career. I kept crossing it, and the crossing became the actual skill.

That's what thirteen years in this gap looks like in practice: the LTE protocol stacks I helped build still carry traffic in a live carrier core network, serving millions of subscribers. The 5G MEC platform I helped extend them into came next. And the edge AI platform I now architect at EPAM (GPU-accelerated computer vision at ≤500ms end-to-end latency across roughly 50 field-deployed devices) is the newest layer on top, not a departure from the rest.

Along the way: provisioning dropped from 15 hours to 45 minutes per device, saving ~710 technician-hours fleet-wide. GPU inference now runs across NVIDIA dGPU, Jetson Orin, and Coral Edge TPU on one portable codebase. V2X microservices hit sub-100ms latency over live 5G. And a multi-agent delivery framework I designed is now in production use across 50+ engineering teams.

Three employers, one client: the same account has kept re-engaging me for eight-plus continuous years, not because of any one employer's name on my badge, but because the work kept needing the specific expertise and experience I'd already built over the years.

If you're building at the intersection of 5G, edge computing, and AI, I'd love to connect. ritwik_jain@live.in

impact

Device provisioning, per unit

before

15h

after

45m

−95%: golden-image builds + AWS SSM agentic flow, fleet-wide

▶view as table

State	Time / device
Before	15 hours
After	45 minutes
Reduction	95%

Interview funnel, EPAM

org

80+

project

10+

~10% hire rate across organizational and project rounds

▶view as table

Stage	Interviews
Organizational	80+
Project	10+
Hire rate	~10%

background

▶EPAM Systems

Sep 2023 – present

Architect & Technical Lead · leads 4–7 engineers · Greater Boston, MA

Delivered ~50 field-deployed devices at ≤500 ms end-to-end video latency by architecting dual-path CV pipelines (third-party + in-house NVIDIA DeepStream/Triton/TAO stack) and building the full Go media layer (RTSP, WebRTC, LL-HLS), resulting in 1 of 3 field trials converted to signed commercial business on a horizontally scalable architecture.
Cut provisioning from 15 hours → 45 minutes per device (95%) by automating golden-image builds and an AWS SSM–based agentic management flow, saving ~710 technician-hours fleet-wide at ~25% lower per-device cost with zero public IP exposure.
Standardized AI-assisted delivery across the org by designing the A-SDLC framework (20 agent personas, 5 coordination protocols, 4,750+ lines across 26 documents), resulting in production adoption by 50+ engineering teams.
Co-authored 4 patent applications (pending). Recipient of EPAM Impact 2025 CEO Award.

▶Leosoft Solutions

Oct 2022 – Sep 2023

Software Developer · Greater Boston, MA

Laid the foundational architecture for the 5G Vision Insight Platform by standing up containerized ThingSpace IoT microservices across MEC and cloud, resulting in a scalable baseline that subsequent engineering phases built directly on top of.
Achieved <100 ms vehicle-to-infrastructure latency by designing V2X microservices as virtual Roadside Units on Cradlepoint 5G routers and Dell IoT Gateways, resulting in validated real-time V2X messaging on commercial 5G hardware.
Eliminated ThingSpace single points of failure by engineering a load-balancing integration across public and private cloud, resulting in higher-availability IoT infrastructure for Verizon enterprise customers.

▶Capgemini (formerly Aricent)

Jan 2013 – Sep 2022

Trainee → Manager / Expert-Lead Technical Engineer · Gurgaon & Greater Boston

Sustained zero client churn across 20+ tier-one telecom clients by owning L3 EPC engineering and resolving 20–40 critical escalations per year (S1AP, NAS, GTP, SIP) over a 9-year tenure.
Delivered EPCLite License Manager in C with 2 bugs across 13K SLOC and zero critical defects across 80+ test cases, by leading end-to-end integration with UDP sockets, resulting in floating feature-based licensing for cloud-deployed EPC.
Expanded EPC hardware coverage to 5 CPU architectures (x86, MIPS, ARM) by porting protocol stacks and optimizing DPDK/Solarflare on NUMA servers, resulting in faster time-to-market across the full client hardware spectrum.
Delivered 10+ technical trainings to client engineering teams, onsite and virtually (EPC architecture, ARM/MIPS cross-compilation, DNS, Lawful Interception), plus multiple internal technical trainings across the org.
Aricent Individual Excellence Award FY2016 (CEO Award). Authored 3 whitepapers; mentored Smart India Hackathon 2017 (Ministry of Civil Aviation).

Progression: Trainee (Jan 2013) · SE (Jul 2013) · SSE (Jul 2015) · Technical Leader (Jul 2018) · Manager (Oct 2021)

awards

2025

EPAM Impact 2025: Impact on Clients and Technology

EPAM Systems · CEO Award

2016

Individual Excellence Award

Aricent, now Capgemini · CEO Award

publications & patents

whitepaper 3 technical whitepapers, authored at Aricent / Capgemini
patent 4 U.S. patent applications, co-authored (pending)
mentorship Smart India Hackathon 2017, Ministry of Civil Aviation

stack

c++

python

kubernetes

nvidia

docker

webrtc

certifications

Claude Certified Architect – Professional

Expires Aug 2027

Claude Certified Developer – Foundations

Expires Aug 2027

Certified Partner Specialist – Gemini Enterprise Agent Development

Expires Feb 2027

AWS Certified AI Practitioner

Expires Aug 2029

Generative AI Leader Certification

Expires Jan 2029

GitHub Copilot Specialist (GH-300)

Expires Jan 2028

Technology Excellence: Solution Architecture

EPAM, Jan 2026

Connected Manager Learning Program

Capgemini / Harvard ManageMentor, Apr 2022

+3 Anthropic, +7 Google Cloud credentials

view on Credly →

open-source

github.com/0x524a/onvif-go

ONVIF client & server for IP camera control: PTZ, streaming, and imaging. Hikvision, Axis, Dahua, Bosch, and any ONVIF-compliant camera.

camera onvif rtsp

▶show example

client, _ := onvif.NewClient(ctx, onvif.Options{
    Host:     "192.168.1.64",
    Username: "admin",
    Password: "***",
})
ptz, _ := client.PTZ()
_ = ptz.AbsoluteMove(ctx, onvif.Vector{
    Pan: 0.5, Tilt: 0.2, Zoom: 0.1,
})

github.com/0x524a/certifier

X.509 certificate authority: RSA, ECDSA, Ed25519. CA, CRL, and OCSP lifecycle in a single library.

security pki tls

▶show example

ca, _ := certifier.NewCA(certifier.CAConfig{
    Algorithm:  certifier.ECDSA,
    KeySize:    256,
    CommonName: "Root CA",
    Validity:   10 * 365 * 24 * time.Hour,
})
leaf, _ := ca.IssueCertificate(certifier.CertificateConfig{
    CommonName: "api.internal",
    DNSNames:   []string{"api.internal"},
    Validity:   365 * 24 * time.Hour,
})

github.com/0x524a/rtspeek

RTSP handshake inspector: OPTIONS/DESCRIBE probing, codec classification, structured JSON output. No live decode required.

rtsp codec cli

github.com/0x524a/metricsd

System, GPU, and HTTP endpoint metrics collection with TLS-native shipping and configurable scrape intervals.

metrics observability tls

github.com/0x524a/NetPulse

Multi-provider network benchmarking: download, upload, and latency across Ookla, Cloudflare, and more.

networking benchmark