Technical Skills

Understanding technical skills is not about becoming an engineer. It is about knowing enough to be effective in your role as a product manager. The trap is trying to master every technical detail — you won't and you don't need to.

Your actual job is to bridge the gap between technology and customer value, not to write code or design databases. But you must speak the language of your engineers well enough to understand their constraints, challenges, and trade-offs.

This lesson teaches you the core technical concepts every PM should grasp, how to assess technical readiness, and how to gain hands-on experience that builds your confidence.

Technical knowledge is context-dependent

How technical you need to be depends on your product and company stage.

If you manage a stable product with mature technology, you can survive with a surface-level understanding. But if you are building a new platform or feature, you must be much closer to the technical details.

Talvinder explains:

"If it's something which is being built up, you cannot avoid being technical. You're making it as you said, right? Product you have to be very much good on the technical side so that you can understand what engineering team needs to do."

In practice, this means knowing the architecture, data flows, and integration points well enough to anticipate risks and communicate clearly.

The essential technical concepts for PMs

Here are the core concepts you should master to be effective without becoming an engineer.

Knowing these concepts helps you understand engineering trade-offs, estimate timelines, and identify risks.

The Planner/Worker pattern in technical architecture

One useful mental model is the Planner/Worker pattern.

Talvinder describes this during a teaching session:

“Notice which option smells like 2015 chatbots – why would static templates break in dynamic pricing?”

Here, the Planner decides what to do next, and the Worker executes the action. This separation helps manage complexity and allows dynamic responses.

Ask yourself: if you add a new tool (like a Price-Drop checker), who decides when to call it — the Planner or the Worker? This helps clarify responsibilities and potential bottlenecks.

AI Memory Systems: short-term vs long-term memory

When working on AI products, understanding memory systems is key.

Talvinder points out:

“Think about Rajesh — his price-sensitivity was learned. That persists across sessions.”

Short-term memory holds immediate context (like the current conversation), while long-term memory stores persistent data (user preferences over time).

In a Redis + Postgres stack, Redis might serve short-term memory for fast access, and Postgres the long-term memory for durability.

Embeddings and vector databases

Embedding models convert data (text, images) into vectors for similarity search.

Talvinder explains:

“A spec sheet converts to a 1536-dim vector. Cosine similarity powers ‘find phones like iPhone 15’.”

Embedding-based search beats hard-coded SQL filters for nuanced queries.

For example, 91mobiles uses embeddings to help users find phones matching complex criteria like “Phones under ₹40k with 120 Hz AMOLED & wireless charging.”

Quality metrics for AI content

Tokens per second impress engineers, but users care about accuracy and relevance.

Talvinder says:

“Wrong specs kill trust. Accuracy ≥ polish.”

Measure content quality by business relevance and factual accuracy, not just speed or volume.

Hands-on system exploration: turning queries into experiments

Technical skills grow fastest by doing.

In the 91mobiles AI workshop, learners run queries, predict outputs, and compare results:

1. Predict silently what the system will return and where it might fail.
2. Run the CLI, measure timings, and note observations.
3. Debrief in groups to compare with the gold standard.

Example queries:

• “Battery capacity of iPhone 15 Pro” → Expect: “4422 mAh (Apple spec, 2024-09) — source link”
• “Compare camera quality Pixel 8 Pro vs Galaxy S24 Ultra” → Expect: Table with sensor size, aperture, lens count, and sources

Common mistakes include returning marketing blurbs without sources or vague paragraphs.

Technology readiness levels (TRL) and prototyping

TRL is a way to grade how mature a technology or feature is.

Talvinder clarifies:

“Docker-compose + staging traffic = operational environment. That’s TRL 6 in practice.”

A TRL 6 prototype means your feature works in a real-like environment under realistic load.

This helps you scope what can be built in a sprint or requires more research.

Integrating architecture knowledge with business context

Technical readiness matters to the CFO because it affects ROI and risk.

Talvinder advises:

“Keep that business hat on; ROI math tomorrow morning depends on TRL.”

Estimating costs and benefits requires knowing how complex the integration is:

• Low: Ready to implement immediately
• Medium: Manageable with planning
• High: Requires significant preparation
• Very High: Need additional expertise

This assessment guides prioritization and resource allocation.

Building confidence through mentorship and collaboration

During workshops, mentors watch for signals:

This system ensures everyone advances and learns effectively.

The honest truth about technical skills for PMs

Talvinder is direct:

“You need to understand those details — not at the depth of an engineer, but enough to communicate clearly and make informed decisions.”

You won’t be asked to write code or explain complex algorithms. But knowing the basics of APIs, databases, and system architecture is essential.

The trap is thinking you must be fully technical before you can be a good PM. That is false.

Slack conversation: Translating model metrics into user impact

This exchange shows how technical metrics must be translated into user experience terms. Your job is to ask the right questions, not just accept numbers.

Meeting scene: Sprint planning with a technical focus

Field exercise: Explore an API integration (time=15 min)

Choose a product or service your company uses that exposes an API (e.g., payment gateway, messaging service).

1. Find the API documentation online.
2. Identify key endpoints and their purposes.
3. Sketch a simple data flow showing how your product interacts with the API.
4. Note any technical constraints or rate limits mentioned.
5. Discuss with an engineer how these constraints might affect your roadmap.

This exercise builds your fluency with technical documentation and integration considerations.

From the field: Talvinder on technical skills

Test yourself: The technical trade-off call

Where to go next

• If you want to learn how to connect technical knowledge to business value: Strategy and Architecture
• If you want to build hands-on skills in AI product development: AI Product Fundamentals
• If you want to improve your stakeholder communication: Stakeholder Management
• If you want to deepen your understanding of user research: User Research Methods

PL alumni now work at Flipkart, Razorpay, Swiggy, PhonePe, and other leading Indian tech companies.