The Role
Lead a 20+ engineer organization across 3–5 pods, setting architecture, reliability, security, compliance, cost strategy, AI evaluation, model routing, integrations, and configuration infrastructure. Build scalable multi-tenant SaaS, media, voice, and data systems while improving unit economics. Own hiring, performance management, engineering standards, incident response, enterprise readiness, and agent governance in a regulated environment.
Summary Generated by Built In
Director, Technology
Scale the engineering organization that turns two products into one platform.
- Function: Product & Technology
- Reports To: CPO / CTO
- Location: Gurugram, onsite
- Work timings: India hours with sustained US overlap
- Experience: 10+ years, including 4+ years leading engineers and at least 2 years managing leads or managers
- Scope: 3 to 5 pods, 20+ engineers, through Technology Leads and Engineering Managers
- Domains open: Vini agents · Studio vision platform · Dealer data and integration platform
- Key Partners: CTO, Head of Product, Post-GTM and Customer Success, Forward Deployed Engineering, Security and Legal, Finance
About Spyne
Spyne builds AI software for US car dealerships. Studio AI turns a phone walk-around of a vehicle into a complete, publish-ready listing. Vini AI runs dealership sales and service engagement across voice, chat and messaging, and writes the outcome back into the dealer’s own systems. We are further building two human-in-the-loop lines, Lot Service and AI Contact Service, which sit on top of them, so the dealer buys the outcome instead of a license.
- Founded: 2020, headquartered in Gurugram with a US field presence
- Scale: More than 3,600 dealerships and automotive brands, reach into 70+ countries, majority of revenue from the United States
- Team: 200+ people, the large majority in India
- Integrations: 70+ live across CRM, DMS, inventory, service scheduling and finance, including a live integration with Tekion Automotive Retail Cloud
- Trust posture: ISO, SOC and GDPR attestations published
- Backing: $25M+ raised from Accel, Vertex Ventures, Storm Ventures and Alteria Capital
What you own
The domain and its architecture
3 to 5 pods through Technology Leads and EMs. You set architecture across the domain, own the seams with adjacent domains, make the build-versus-buy calls on model providers, telephony, eval tooling and memory infrastructure, and decide what gets deprecated. You are expected to say no to a second way of doing something that already exists.
Reliability and unit cost at scale
SLOs, error budgets, on-call health, incident command, and written postmortems that produce a change rather than a document. You own the cloud and inference bill for your domain, and you are expected to bring per-unit cost down while volume goes up.
Security, compliance and enterprise readiness
Threat modeling in design, tenant isolation, secrets and access, PII across transcripts and traces, vulnerability and dependency management, audit evidence. You own your domain’s part of the security pack that lets a Top 150 dealer group reach a contract instead of stalling at the questionnaire. You will sit in front of a group CIO. In a regulated contact environment, consent, recording and TCPA are design inputs.
The hiring engine
Interview loop design, calibration of interviewers, a written bar, and a funnel you measure: pass-through, time to close, and 12-month retention of your own hires. At our growth rate you will run [15 to 25] hires a year across the domain. You close senior candidates yourself.
Performance management
Levels and expectations written down. Feedback within days, not at cycle end. Calibration with peer directors. You are expected to move on underperformance inside a quarter, and to develop two leads into people who could replace you. We run a candor-first culture. Coasting is not managed around, it is addressed.
Agentic AI: key requirements
At this level we are not testing whether you can build an agent. We are testing whether you can build an organization that ships agents which stay reliable while the models underneath them change every few months.
- You set the eval standard for the domain and you enforce it. Which metrics are real and which are vanity, what constitutes a golden set, what runs in CI, what blocks a release. You can explain to the CEO why a 3-point movement is noise.
- You own model strategy as an economic decision, not a preference. Routing between small and large models, open and closed, per task. Caching, distillation and fine-tuning where they earn their cost. Provider concentration risk and a tested fallback path. You can produce a cost-per-interaction curve for the last two quarters and explain every step in it.
- You know where the moat is and where it is not. Voice quality is not defensible, because every vendor improves when the foundation models improve. What is defensible is write-back depth into systems that resist integration, service-side workflow the incumbents never bothered to model, and accountability for a booked and kept appointment rather than a handled call. Your architecture should reflect that.
- The configuration surface is product infrastructure and you treat it that way. Thousands of dealers each wanting a different agent cannot mean per-dealer code. Every hour a deployment engineer spends on a customer is a defect in the configuration surface. You are measured on how much per-customer work converts into product.
- Multi-agent orchestration over a shared record, where the interesting problems are handoff, state, partial failure and who is allowed to write. Not a diagram of boxes talking to each other.
- Agent safety and governance in a regulated environment. The agent speaks on the dealer’s phone line, so a bad call is attributed to the dealer rather than to us. Brand-voice governance, escalation policy, consent, recording, TCPA, PII redaction, and auditability of every automated action taken inside a customer’s system.
- AI-native engineering practice inside your own org. Coding agents, generated tests, agent-assisted review and eval authoring, used with judgment about where they help and where they create rework. You are expected to have a view on how your team’s shipping rate changed because of it, with evidence.
Systems and architecture depth
- Multi-tenant SaaS at thousands of tenants, high-throughput media pipelines, and realtime voice at concurrency, with the cost discipline all three imply.
- Integration architecture against hostile surfaces: certified vendor programs such as Fortellis, RCI, Dealertrack and Tekion APC, rate limits, partial and stale data, no usable test environment, and a partner with a commercial interest in the integration being difficult. Sequencing certifications so a deal is never waiting on one.
- Data architecture for a shared vehicle and customer record across products, including identity resolution, event sourcing, and roll-up reporting across a multi-rooftop group.
- You can read a design doc and find the failure that shows up at 10x volume.
This role is not for you if
- You manage through status reports and cannot evaluate a design yourself.
- Your AI exposure is governance frameworks and vendor evaluation.
- You need a stable org chart. Ours is being built while the company scales.
- You are uncomfortable being told directly that something you shipped is not good enough, or telling someone else.
Skills Required
- 10+ years of engineering experience
- 4+ years leading engineers
- At least 2 years managing technology leads or engineering managers
- Experience managing 3–5 pods and 20+ engineers through leads and managers
- Ability to set architecture and make build-versus-buy decisions
- Experience with SLOs, error budgets, on-call health, incident command, and postmortems
- Experience owning cloud and AI inference costs at scale
- Experience with security, compliance, tenant isolation, PII, vulnerability management, and audit evidence
- Experience designing evaluation standards, golden sets, CI checks, and release gates for AI agents
- Experience with model routing, caching, distillation, fine-tuning, provider fallback, and cost analysis
- Experience with multi-agent orchestration, shared state, handoffs, partial failure, and write permissions
- Experience with agent safety, consent, recording, TCPA, PII redaction, and auditability
- Experience with multi-tenant SaaS, high-throughput media pipelines, and real-time voice systems
- Experience integrating with Fortellis, RCI, Dealertrack, Tekion APC, or similarly difficult vendor platforms
- Experience with shared-record data architecture, identity resolution, event sourcing, and multi-rooftop reporting
- Ability to evaluate technical designs and identify failures at increased scale
- Ability to design engineering interview loops, calibrate interviewers, and hire senior candidates
- Ability to establish levels, expectations, feedback, calibration, and performance management practices
- Willingness to work onsite in Gurugram during India hours with sustained US overlap
Am I A Good Fit?
Get Personalized Job Insights.
Our AI-powered fit analysis compares your resume with a job listing so you know if your skills & experience align.
Success! Refresh the page to see how your skills align with this role.
The Company
What We Do
One of the biggest pain points for any online seller is to showcase their products in the best possible manner to grab customer attention and increase conversions. Spyne is transforming the way businesses create their catalog using state-of-the-art AI technology. No studio. No photography skills. No complex processes. Just Spyne to create stunning catalogs that drive 40% better conversions. We are building first-of-its-kind AI that would transform the way businesses will create high-impact catalogs without the need for a physical studio. We are building computer vision and AI technology to automate image processing workflows. This will help any business or marketplace shoot, edit, and publish product visuals at scale, at a fraction of the cost, and time. We are helping large e-commerce marketplaces in the automotive, food, fashion, real estate, and retail industry create and upgrade high-quality visual catalogs at scale. Founded in 2018, Spyne is headquartered in Gurugram, India serving 80+ customers across 15+ countries including Amazon India, Karvi, OLACars, and The Luxury Closet Dubai. We are backed by Accel Partners, Storm Ventures, and other marquee investors.
.png){kind=logo}
