🎯 When Specialisation Wins

AI/ML Engineering: Deep specialisation in LLMs, MLOps, or recommendation systems commands 30–60% premiums over equivalent-level generalists.

Security Engineering: AppSec + cloud security at senior levels is extremely scarce — specialist premium is consistent.

Embedded / VLSI: Non-transferable domain expertise. Specialists are always in short supply.

Data Engineering (Spark/Kafka/dbt): Deep stack knowledge is valued; generalist SQL analysts don't compete.

🌐 When Generalism Wins

Early-stage startups (Seed–Series A): Full-stack engineers who can ship independently and wear multiple hats are more valuable than specialists.

Engineering leadership roles: EMs and Directors need breadth — depth in one stack is a handicap at senior management levels.

Platform / DevOps roles: Cross-domain knowledge (cloud + containers + CI/CD + networking) is the product.

Technical CTO (early-stage): Needs to make decisions across the entire stack.

🏭 GCC / MNC (Global Capability Centres)

Scale: India has 1,700+ GCCs employing 1.5M+ engineers. Growing 15–20% annually.

Engineering character: Feature development for global products. Process-heavy. FAANG-adjacent compensation. Strong in cloud, data, and QA.

Career pattern: Predictable ladder. Technology risk is low. Best for: IC depth in a specific domain (Java, AWS, QA automation, data engineering).

Limitation: Product ownership is rare. Engineering leadership ceiling is real — strategic decisions stay at HQ.

🚀 Indian Product Companies

Examples: Flipkart, Swiggy, Zepto, Razorpay, CRED, PhonePe, Zomato, Urban Company.

Engineering character: Full product ownership. High velocity. Engineering leaders make real product decisions. Best engineering culture in India.

Career pattern: Fast — SDE2 to Staff in 3–4 years at the best companies. High ownership, high ambiguity.

Compensation: Stock-heavy. Cash vs GCC may be comparable; total comp (ESOP) often exceeds at senior levels.

💡 Startups (Seed to Series C)

Engineering character: High ambiguity, high ownership. Build everything from scratch. Technical debt is real. Product-engineering boundary is blurred.

Career pattern: Non-linear. An SDE2 at Series A can become a Director at Series C. Risk is high; reward is high.

What to look for: Engineers who thrive in ambiguity, have built things end-to-end, and can make technology decisions without an architecture review board.

Limitation: Engineering process maturity is low early-stage — candidates from large companies struggle.

Senior Engineer (5–8 yrs)

Owns: Complete features and system components. Mentors junior engineers. Makes technology choices within a team context.

Green flag: "I designed this system — here is the trade-off I made between consistency and availability and why."

Red flag: Senior engineers who code well but can't communicate design decisions to others — technically strong but not yet leadership-ready.

Staff / Principal Engineer

Owns: Architecture across multiple teams. Technical direction for a product area or platform. Multiplies other engineers.

Green flag: "I wrote this RFC that changed how three teams thought about data consistency — here's the problem, my proposal, and how I got alignment."

Red flag: Engineers who claim Staff level but can't show cross-team impact — deep technical work that only benefits their own team is Senior, not Staff.

Engineering Manager

Owns: A team's delivery, health, and growth. Partners with PM. Represents engineering in cross-functional discussions.

Green flag: "The hardest situation I've managed was an underperforming engineer — here's how I handled it, what changed, and what I learned."

Red flag: EMs who still position themselves as the best coder on the team — they haven't made the management transition.

VP Engineering / CTO

Owns: The engineering organisation, culture, and technology direction. A business partner who happens to run engineering.

Green flag: "Here's the biggest engineering culture problem I diagnosed — what was broken, what I changed, and how I measured whether it worked."

Red flag: VPs/CTOs who can only talk about technology decisions — at this level, the org design IS the primary product.

Staff Engineer / Principal Engineer

The IC leadership track is poorly understood even within engineering. Most companies promote people into Staff without a clear charter. The supply of genuine Staff Engineers (cross-team impact, RFC culture, architectural vision) is extremely thin.

AI / ML Engineering (Senior+)

Everyone claims ML experience post-ChatGPT. Genuine production ML Engineers — who've deployed models, built feature stores, monitored drift, and managed retraining pipelines — are scarce and expensive. The gap between a notebook ML analyst and a production ML engineer is enormous.

CTO (Early-Stage Startup)

Early-stage CTO must be chief architect, first engineer, technical hiring manager, and technology communicator to investors simultaneously. This combination is rare. Most engineering leaders are strong in one or two of these — rarely all four.

Security Engineering (AppSec)

Application security engineers who write production-quality code AND understand offensive security AND can build security culture are extremely rare. Most security roles are filled by compliance-heavy GRC profiles that don't meet engineering security requirements.

DevOps / Platform Engineering (Staff+)

Senior platform engineers who understand developer experience, cloud cost, reliability, AND security simultaneously are always in demand and rarely available. Most DevOps profiles are tool-heavy without systems thinking.

Embedded / Firmware (Automotive ASIL)

ISO 26262 functional safety experience combined with AUTOSAR architecture expertise is extremely scarce in India. Automotive ECU engineers are concentrated in Pune and are heavily poached by GCCs of German OEMs.

For Backend / System Design

"Design a rate limiter that needs to handle 100K requests/sec with sub-millisecond latency — walk me through your approach, data structures, and trade-offs." Tests: distributed systems thinking, not just API knowledge.

For ML Engineers

"Walk me through a model you've deployed to production — serving infrastructure, latency requirement, how you monitored it post-deployment, and how you handled data drift." Tests: production ML depth, not just training.

For Staff / Principal Engineers

"Tell me about a technical RFC you've written that changed how multiple teams work — what was the problem, your proposal, and how did you build consensus across disagreeing engineers?" Tests: cross-team influence and communication.

For Engineering Managers

"Walk me through the most difficult underperformance situation you've managed — what was the issue, how long did you give it, what interventions did you make, and what was the outcome?" Tests: people management reality, not theory.

For GenAI / LLM Engineers

"Walk me through a production RAG system you've built — what was the chunking strategy, embedding model, retrieval approach, evaluation methodology, and how did you handle hallucinations?" Tests: production GenAI depth vs tutorial exposure.

For VP Engineering / CTO

"What is the biggest technology or engineering org mistake you've made as a leader — what happened, why did you make that call, and what would you do differently?" Tests: self-awareness, leadership maturity, and learning velocity.

📷 Consumer Internet (Flipkart, Swiggy, Zomato, CRED)

Engineering priority: Scale, reliability, latency, data engineering, ML/personalisation, mobile engineering

Key signals: Experience with high-QPS systems, on-call culture, DORA metrics awareness, distributed systems fluency

Red flag: Engineers who can't articulate how they've designed for failure — in consumer internet, systems fail at scale constantly

🏭 GCC / MNC (Microsoft, Google, Amazon, Goldman, JP Morgan Tech)

Engineering priority: Feature development for global products, quality engineering, cloud, data platform, security

Key signals: Depth in one domain, process discipline, cross-timezone collaboration, documentation culture

Red flag: GCC engineers applying to early-stage startups — the ambiguity tolerance and ownership expectation gap is real and large

💳 Fintech (Razorpay, PhonePe, Paytm, Groww, Zerodha)

Engineering priority: Payment rails, fraud engineering, compliance systems, reliability (99.99% uptime is table stakes), real-time data

Key signals: UPI / payment system experience, strong on reliability and SRE culture, understanding of regulatory constraints

Red flag: Engineers who treat financial engineering like regular consumer apps — the stakes (real money, regulatory risk) require a different mindset

💻 B2B SaaS (Freshworks, Chargebee, Zoho, Postman)

Engineering priority: Multi-tenancy, API-first architecture, integrations ecosystem, security (SOC 2), performance at enterprise scale

Key signals: Multi-tenant architecture experience, API design quality, understanding of SLAs and enterprise customers

Red flag: Consumer internet engineers for B2B SaaS without understanding the different customer expectation — enterprise wants stability, not velocity

🚀 Deep-Tech / Hardware (automotive, semiconductor, aerospace)

Engineering priority: Embedded systems, VLSI, real-time OS, functional safety, hardware-software co-design

Key signals: Domain certification (ISO 26262 for automotive, DO-178C for aerospace), hardware debug tools experience, C/C++ embedded depth

Red flag: Embedded engineers without hardware debugging experience — the ability to read a schematic and use an oscilloscope is table stakes

🏠 HealthTech (Practo, Niramai, mFine, PharmEasy)

Engineering priority: Data privacy (DPDP Act, HIPAA for global), reliability, EHR integration, ML for diagnostics, regulatory compliance

Key signals: Understanding of health data sensitivity, FHIR/HL7 standards (for clinical systems), security-first engineering mindset

Red flag: HealthTech companies need engineers who understand why their data handling is different from a food delivery app — privacy is non-negotiable

SDE 1 / Junior Engineer (0–2 yrs)

GCC / Service company: ₹6–15L

Product startup: ₹10–20L

Top product company (Flipkart, PhonePe): ₹18–35L

FAANG India: ₹30–60L

IIT premium at SDE1: ₹5–15L above median

SDE 2 / Mid-Level Engineer (2–5 yrs)

GCC / Service company: ₹15–30L

Product startup: ₹22–45L

Top product company: ₹35–70L

FAANG India: ₹50–100L

AI/ML SDE2 premium: 30–50% above median

Senior Engineer (5–8 yrs)

GCC / Service company: ₹25–50L

Product startup: ₹40–80L

Top product company: ₹60–130L

FAANG India: ₹80–200L

Security / DevOps specialisation: ₹10–30L premium

Staff / Principal Engineer (8–14 yrs)

GCC / Service company: ₹40–80L

Product startup: ₹70–150L + ESOP

Top product company: ₹100–250L

FAANG India: ₹150–400L+

ML Staff Engineer: can reach ₹200–500L at FAANG

Engineering Manager / Director (8–15 yrs)

EM (GCC): ₹30–70L

EM (Product startup): ₹50–120L + ESOP

Director (GCC): ₹60–130L

Director (Product company): ₹100–250L

Management track generally earns less than IC Staff at top companies

VP Engineering / CTO (15+ yrs)

Startup (Seed/A): ₹40–80L + 1–4% ESOP

Mid-size company (Series B/C): ₹80–200L + ESOP

Large product company: ₹200L–1Cr+

MNC India CTO: ₹200L–3Cr

Early-stage CTO: cash below market; ESOP is the bet

Scenario 1: Senior vs Staff Engineer Confusion

Client wants a "Staff Engineer" for their platform team. Budget: ₹80L. They describe the role: "Own the backend of our payments service, write good code, review others." You find a genuinely strong Senior Engineer at ₹65L and a Staff Engineer at a FAANG for ₹200L.

The move: Clarify what Staff actually means. A Staff Engineer owns cross-team technical direction, writes RFCs, and multiplies engineers beyond their immediate team. What the client described is a Senior Engineer role. Don't let the client overpay for the wrong level, or underpay and lose the Staff candidate. Re-brief the client on scope before proceeding.

Scenario 2: The AI/ML Hype Filter

Client needs an "ML Engineer to build AI features." 60 CVs received, all claim "ML experience." Budget: ₹40–60L. Most profiles: data analysts who've used scikit-learn in notebooks. A few: actual production ML engineers at e-commerce or fintech companies.

The move: Apply a three-question filter before sending any profile: 1) "Have you deployed a model to production?" 2) "What was the serving infrastructure and latency SLA?" 3) "How did you monitor for data drift?" Analysts who've only done offline experiments will fail all three. Share only profiles that pass — and explain the filter to your client so they trust your shortlist.

Scenario 3: GCC Engineer for a Startup

Series B startup (₹120Cr ARR, 45 engineers) needs a Director of Engineering. Strong candidate: 14 years experience, last 6 years at a top GCC managing a team of 25 engineers. Excellent process discipline, people skills, strong tech background. Never worked at a company smaller than 5,000 people.

The move: Probe the ambiguity tolerance explicitly: "In your current role, when you need a new tool or process, what does the approval and rollout look like?" Then ask: "In a 45-person startup, you'd make that decision yourself by Tuesday. How does that feel?" If they lean in — worth presenting. If they look uncomfortable — flag it to the client as a genuine risk.

Scenario 4: The Early-Stage CTO Mandate

Seed-stage startup (20 people, 5 engineers, pre-PMF) wants to hire a CTO. Budget: ₹60–80L cash + 2% ESOP. Looking for a "technical visionary."

The move: Be direct with the founder: at Seed, the CTO IS the engineering team. They need someone who can write code today AND make architecture decisions for year 2. Ask: "How important is it that your CTO ships code in the first 6 months?" Budget reality: ₹60–80L cash will not attract a late-stage VP Engineering. Focus on ambitious Senior/Staff Engineers ready for their first leadership role.

Scenario 5: The Security Engineering Mismatch

Fintech client (₹500Cr ARR, PCI-DSS compliance required) needs a "Head of Security." Profiles: (A) CISO from a large bank — deep GRC/compliance, no engineering background. (B) AppSec Engineer — 8 years, writes code, OSCP certified, has built bug bounty programmes. (C) Security architect — 12 years, designed zero-trust architectures, comfortable with both code and governance.

The move: Profile A is a Compliance Head, not an engineering security leader. Present Profiles B and C with the framing: "B is strong on engineering depth and offensive security. C has broader architecture and governance." Let the client choose based on what they're missing most.

Scenario 6: The GenAI Title Inflation

Consumer internet client wants a "Senior GenAI Engineer." Budget: ₹60–90L. In screening: 2 candidates have integrated ChatGPT API via LangChain tutorial; 1 has built a production RAG system handling 500K daily queries with custom evaluation; 1 has fine-tuned models and built agent workflows.

The move: Title inflation in GenAI is extreme. Only share candidates 3 and 4. For candidate 3, ask: "Walk me through your evaluation framework — how do you measure RAG quality beyond basic accuracy?" For candidate 4, ask: "What did fine-tuning add that prompting alone couldn't give you?" If they can't answer in technical detail, they're a power user, not an engineer.

SLA / SLO / SLI

SLI (Service Level Indicator) = measurement (e.g., latency). SLO (Service Level Objective) = internal target (e.g., p99 latency < 200ms). SLA (Service Level Agreement) = external commitment to customers. SREs manage error budgets based on SLOs. Ask any platform or backend candidate: "What was your team's SLO for your primary service?"

CI/CD

Continuous Integration (CI) = automatically testing code changes. Continuous Delivery/Deployment (CD) = automatically releasing tested changes to production. A team without CI/CD ships slowly and with high risk. Ask any DevOps or senior backend candidate: "Describe your current deployment pipeline — how long from commit to production?" Elite teams: under 30 minutes.

Infrastructure as Code (IaC)

Managing cloud infrastructure through code (Terraform, Pulumi) rather than clicking in a console. Enables reproducibility, version control, and disaster recovery. A DevOps engineer without IaC experience is a cloud admin, not a platform engineer. Ask: "What's your Terraform module structure for a multi-environment AWS setup?"

DORA Metrics

Four metrics that measure software delivery performance: 1) Deployment Frequency, 2) Lead Time for Changes, 3) Change Failure Rate, 4) Time to Restore. Elite teams: deploy multiple times/day, lead time < 1 day. Good DevOps and EM candidates should know and quote these without prompting.

LLM vs Traditional ML

Traditional ML: train a model on specific data for a specific task (churn prediction, fraud detection). LLMs: large pre-trained models that can be prompted or fine-tuned for many tasks. They're complementary — not interchangeable. An ML Engineer trained on traditional ML is NOT automatically qualified for GenAI/LLM engineering. Different stack, different failure modes, different debugging approaches.

SRE vs DevOps

DevOps = culture + practices for faster, more reliable software delivery. SRE = Google's engineering discipline for reliability — SLOs, error budgets, toil elimination, production systems management. SREs typically come from software engineering backgrounds; DevOps practitioners from operations. Both are scarce. Neither is a synonym for "sysadmin who knows Docker."