Pusaram AI Academy
Lessons
Four complete draft lessons you can read today, and an outline for every other unit in the curriculum.
Ready to read: the AI Awareness Weekend
Four lessons, about an hour each, with a worked example, a quiz and a practical task. They are drafts: complete, but not yet tried with learners or reviewed by an independent specialist.
- O01
What AI can help with and where judgment stays with you
Given an everyday task, distinguish a fixed rule, a prediction, a generated draft, and a decision that needs human responsibility. No programming is required.
- O02
Give clear instructions and improve a useful draft
Write a request with a goal, relevant context, constraints, and an output format; then improve a draft using specific feedback. You can type, dictate, or work with a partner.
- O03
Check claims, calculations, and missing information
Use a short evidence packet and an independent calculation to identify an unsupported answer and write a justified correction.
- O04
Protect information and choose your next useful project
Choose proportionate information sharing, identify who has authority to act, and write a small next-step project with a review plan.
Outlines for the longer programs (156 units)
These are plans, not lessons: each shows the planned outcome, time and selected reading. They are published so you can see where the curriculum is going.
D00 Digital confidence bridge
- D00-U01 Files, folders, and downloads Create, find, rename, and move a synthetic practice file
- D00-U02 Browsers, accounts, and permissions Explain what an upload and a permission grant allow
- D00-U03 Typing, voice, and accessible interaction Complete a task using a comfortable input method
- D00-U04 Tables, sharing, and basic troubleshooting Share a practice artifact appropriately and recover a misplaced file
A10 AI literacy and responsible judgment
- A10-U01 AI, machine learning, and generative systems Classify familiar tasks as rules, prediction, retrieval, or generation
- A10-U02 How AI developed and why it matters Connect major approaches to a current strength and limitation
- A10-U03 What an AI answer can and cannot establish Identify unsupported statements and missing evidence
- A10-U04 Privacy, fairness, and human responsibility Choose a safe input and identify the accountable reviewer
A20 Practical prompting and verification
- A20-U01 Goals, context, constraints, and output format Rewrite an unclear request into an assessable instruction
- A20-U02 Examples, iteration, and context selection Improve a draft using relevant context and a comparison
- A20-U03 Evidence, calculations, and uncertainty Check one factual claim and one calculation independently
- A20-U04 Repeatable task recipes and personal rules Document a reusable recipe with checks and limits
P10 Everyday planning and personal administration
- P10-U01 Planning with real constraints Create a weekly plan that respects time, budget, and priorities
- P10-U02 Comparisons and decision tables Compare alternatives using explicit criteria and missing information
- P10-U03 Letters, forms, and document understanding Draft a clear message without disclosing sensitive identifiers
- P10-U04 Family and community coordination Prepare an accessible event or volunteer coordination pack
P20 Learning, creativity, and responsible media
- P20-U01 Learning plans and active practice Build a learning routine that requires recall and independent work
- P20-U02 Writing, translation, and cultural context Revise a bilingual message and inspect changes in meaning
- P20-U03 Images, audio, and synthetic-media literacy Create a media brief and explain consent and attribution needs
- P20-U04 Personal project and reflection Deliver a useful artifact and explain what was checked
W10 Professional productivity and communication
- W10-U01 Emails, reports, and presentations Transform raw notes into a concise audience-appropriate document
- W10-U02 Meetings and action management Separate decisions, actions, owners, and unresolved points
- W10-U03 Spreadsheets and analysis assistance Check an AI-generated formula on known inputs
- W10-U04 Research and synthesis at work Produce a sourced brief with uncertainty and contradictory evidence
W20 Workplace workflow redesign
- W20-U01 Map the process before choosing a tool Document inputs, decisions, handoffs, and exceptions
- W20-U02 Measure a useful baseline Record time, errors, and review effort for comparable tasks
- W20-U03 Choose AI, rules, or human work Allocate steps according to uncertainty and consequences
- W20-U04 Pilot, feedback, and operating procedures Write a bounded pilot plan and a clear stopping rule
C10 Documents, knowledge, and multimodal work
- C10-U01 Document structure and extraction Check text and table extraction against an original page
- C10-U02 Knowledge sources and freshness Resolve an expired versus current policy example
- C10-U03 Images, speech, and transcription at work Identify modality-specific errors and a text alternative
- C10-U04 Permission-aware collaboration Describe who may see a source, answer, and exported artifact
V10 Quality, privacy, and responsible professional use
- V10-U01 Quality rubrics and representative cases Define success and failure for a recurring task
- V10-U02 Human review and calibrated confidence Explain why confidence language is not correctness evidence
- V10-U03 Data handling, intellectual property, and disclosure Classify synthetic examples by allowed handling
- V10-U04 Failure reporting and improvement Turn an observed failure into a regression test and action
AU10 No-code and low-code automation foundations
- AU10-U01 Triggers, actions, and process state Draw a workflow with clear start, finish, and exception states
- AU10-U02 Data mappings and conditions Validate required fields before calling an AI step
- AU10-U03 Connections, credentials, and authority Configure or document minimum required access
- AU10-U04 Human approval and exception queues Route uncertain or consequential actions to review
- AU10-U05 First end-to-end automation Demonstrate a simulated intake-to-draft workflow
- AU10-U06 Testing and useful monitoring Record outcome, duration, and failure cause for test events
AU20 Dependable automation and operations
- AU20-U01 Retries, timeouts, and duplicate events Show that repeating a test event does not duplicate its side effect
- AU20-U02 Schedules, webhooks, and changing data Trace a scheduled or event-driven update to a final record
- AU20-U03 Logs, alerts, and spending limits Detect a repeated failure and stop excessive processing
- AU20-U04 Permission and adversarial tests Contain an untrusted instruction inside an incoming document
- AU20-U05 Pilot economics and handover Compare total effort including reviews and maintenance
- AU20-U06 Recovery drill and capstone Recover a failed run and hand over the operating instructions
T10 Transformation strategy and value design
- T10-U01 Opportunity discovery and process diagnosis Identify where a process problem actually originates
- T10-U02 Value, feasibility, and risk prioritization Rank opportunities using stated criteria and uncertainty
- T10-U03 Build, buy, and vendor assessment Compare options using portability, data, and operating needs
- T10-U04 Business case and sensitivity analysis Calculate a range of benefits and costs without invented returns
- T10-U05 Governance and accountable ownership Assign decisions, reviewers, and escalation paths
- T10-U06 Pilot portfolio and roadmap Sequence bounded pilots with measurable entry and exit conditions
T20 Adoption, change, and responsible scale
- T20-U01 Stakeholders and adoption barriers Distinguish skill, trust, incentives, and process barriers
- T20-U02 Role redesign and learning plans Create role-specific training without unsupported job predictions
- T20-U03 Operating model and governance forums Define who owns data, changes, incidents, and outcomes
- T20-U04 Pilot evaluation and rollout decision Compare observed benefit with baseline and residual risk
- T20-U05 Scaling, procurement, and continuous review Define expansion conditions and vendor exit arrangements
- T20-U06 Transformation defense and handover Defend a roadmap against a skeptical stakeholder review
D10 Python and computational problem solving
- D10-U01 Environment, files, and program execution Run a script in an isolated documented environment
- D10-U02 Values, collections, and control flow Transform records with conditions and loops
- D10-U03 Functions, modules, and error handling Separate a task into testable functions
- D10-U04 Data files, tables, and validation Parse CSV and JSON and reject malformed inputs
- D10-U05 Debugging, tests, and AI-assisted coding Repair an unfamiliar seeded defect and explain the fix
- D10-U06 Reusable program and independent challenge Deliver a reproducible script on a new dataset
D20 Data literacy, SQL, and knowledge lifecycle
- D20-U01 Data models and quality Document units of observation, missingness, duplicates, and provenance
- D20-U02 SQL and relational thinking Query related tables with keys, joins, and aggregation
- D20-U03 Documents, parsing, OCR, and tables Validate extracted fields against source evidence
- D20-U04 Pipelines, versions, and effective dates Trace an input through transformation and correction
- D20-U05 Permissions, retention, and deletion Demonstrate removal or access restriction in downstream results
- D20-U06 Dataset card and ingestion project Reproduce an ingestion run and explain remaining limitations
D30 Software engineering and reproducible delivery
- D30-U01 Git, environments, and dependencies Reproduce a tagged version using documented dependencies
- D30-U02 Design, typing, and contracts Define module boundaries and validate structured inputs
- D30-U03 Tests, mocks, and integration checks Test expected, malformed, and unavailable-service cases
- D30-U04 Code review, documentation, and change Explain a design choice and implement a small change request
L10 Language models and AI application patterns
- L10-U01 Tokens, embeddings, and transformer intuition Explain context and token constraints without claiming comprehension
- L10-U02 Training, post-training, and model selection Choose a model interface for a declared task and budget
- L10-U03 Prompt and context engineering Separate trusted instructions from untrusted source content
- L10-U04 Structured output and validation Reject a schema-valid but unsupported field when evidence is missing
- L10-U05 Model interfaces, cost, and streaming Handle errors and record task-level usage
- L10-U06 Comparison and application lab Compare two configurations on fixed examples
R10 Retrieval and grounded knowledge systems
- R10-U01 Lexical search and relevance labels Create relevance judgments and a keyword baseline
- R10-U02 Embeddings, chunks, and vector search Compare chunk choices against known questions
- R10-U03 Hybrid retrieval and reranking Test whether added retrieval complexity improves a fixed task
- R10-U04 Citations, conflicts, and abstention Distinguish relevant passages from claims they actually support
- R10-U05 Retrieval metrics and error diagnosis Separate retrieval miss from unsupported generation
- R10-U06 Permission and knowledge-update lab Demonstrate source updates, restricted access, and deletion
E10 Evaluation, experiments, and observability
- E10-U01 Task metrics, labels, and baselines Define success and a simpler comparator for the chosen system
- E10-U02 Splits, uncertainty, and representative slices Keep tuning separate from final evaluation
- E10-U03 Human rubrics and automated judges Calibrate judgments using human-reviewed examples
- E10-U04 Traces, versions, and regression checks Trace a failure to data, prompt, model, tool, or code
- E10-U05 Comparative experiment and limitations Report raw counts, tradeoffs, and sample limitations
- E10-U06 Evaluation report and release recommendation Make a supported go/no-go recommendation
B10 Backend APIs and integrated services
- B10-U01 HTTP, APIs, and validated contracts Handle status codes, pagination, and malformed requests
- B10-U02 Authentication, permissions, and secrets Separate identity from permission and protect credentials
- B10-U03 Databases, transactions, and storage Persist state without inconsistent partial updates
- B10-U04 Async work, queues, and retries Handle slow work with timeouts and idempotency
- B10-U05 Interface, usability, and accessible feedback Provide success, waiting, correction, and failure states
- B10-U06 End-to-end integration lab Demonstrate a recoverable workflow with mock external systems
AG10 Agents, tools, memory, and bounded execution
- AG10-U01 Workflows versus autonomous loops Choose the simplest adequate execution design
- AG10-U02 Tool schemas and authority Validate arguments and separate read from write capabilities
- AG10-U03 State, context, and memory Define what is stored, for how long, and with whose authority
- AG10-U04 Checkpoints, stop conditions, and approval Resume safely and stop repeated or over-budget actions
- AG10-U05 Protocols and orchestration patterns Explain protocol boundaries and a justified handoff
- AG10-U06 Agent failure and comparison lab Test malformed calls, untrusted tool output, and duplicate actions
S10 AI security, privacy, and governance engineering
- S10-U01 Assets, trust boundaries, and threat scenarios Identify credible entry points and consequences
- S10-U02 Prompt injection, leakage, and tool misuse Test containment in a controlled synthetic environment
- S10-U03 Access, supply chains, and data lifecycle Verify minimum permissions and dependency/source trust
- S10-U04 Governance, incidents, and residual risk Assign owners and document unresolved limitations
O10 Deployment, reliability, and operational economics
- O10-U01 Packaging and environment configuration Rebuild a declared release from documented artifacts
- O10-U02 CI and evaluation release gates Block a deliberately broken change
- O10-U03 Health, logs, alerts, and service objectives Detect a seeded outage without exposing sensitive logs
- O10-U04 Backpressure, rate limits, and graceful failure Protect the service when requests or dependencies fail
- O10-U05 Rollbacks, backups, and incident drills Restore a prior working version and required data
- O10-U06 Cost, capacity, and handover Estimate cost per successful task and assign operating ownership
MA10 Mathematics and statistics for model builders
- MA10-U01 Linear algebra and vector operations Explain shape, dot product, distance, and matrix operations
- MA10-U02 Probability and conditional reasoning Compute a conditional probability and discuss base rates
- MA10-U03 Statistics and sampling Explain variability and how sampling affects a conclusion
- MA10-U04 Calculus and optimization intuition Relate a gradient to reducing a simple loss
- MA10-U05 Experimental design and uncertainty Design a comparison with an appropriate split and limitation
- MA10-U06 Numerical stability and interpretation Diagnose an implausible calculation or model statistic
ML10 Predictive machine learning and responsible validation
- ML10-U01 Problem, target, and baseline Choose a target and comparator that match a real decision
- ML10-U02 Preprocessing and leakage control Fit transformations using training data only
- ML10-U03 Regression, classification, and trees Compare appropriate simple models
- ML10-U04 Unsupervised learning and anomaly methods Explain when labels are absent and evaluation is difficult
- ML10-U05 Imbalance, calibration, and explanation Analyze decision errors across relevant slices
- ML10-U06 Time-aware evaluation and final report Protect a held-out evaluation and report limits
DL10 Neural learning and multimodal foundations
- DL10-U01 Neurons, layers, loss, and training Explain a small forward pass and learning update
- DL10-U02 Backpropagation and optimization Inspect training curves and diagnose a learning failure
- DL10-U03 Overfitting, regularization, and transfer Compare training and validation behavior
- DL10-U04 Vision, speech, and representation learning Compare modality-specific inputs, labels, and errors
- DL10-U05 Attention and transformer architecture Trace shapes and information flow in a simplified example
- DL10-U06 Small-model experiment and model card Compare against a simpler baseline with resource limits
AD10 Adaptation, compression, and advanced-model judgment
- AD10-U01 Prompting, retrieval, and tuning decisions Choose an intervention based on observed errors
- AD10-U02 Data curation and adaptation evaluation Separate adaptation data from evaluation cases
- AD10-U03 Fine-tuning, LoRA, and alignment overview Explain different objectives and resource requirements
- AD10-U04 Quantization, distillation, and serving Compare efficiency gains against quality losses
- AD10-U05 Resource-feasible adaptation lab Run a small or simulated-labeled experiment with clear evidence limits
- AD10-U06 Change, model card, and go/no-go decision Recommend adoption or rejection using measured tradeoffs
MD10 Model lifecycle and MLOps
- MD10-U01 Datasets, features, and experiment records Reproduce a training run with traceable inputs
- MD10-U02 Model registries and artifact lineage Link a model version to data, code, and evaluation
- MD10-U03 Serving, batch scoring, and monitoring Choose a scoring pattern and monitor outcomes
- MD10-U04 Drift, labels, and retraining decisions Distinguish distribution change from measured quality loss
- MD10-U05 Release gates and rollback Reject an inferior candidate and restore the previous model
- MD10-U06 Lifecycle handover and review Assign owners for data, model, incidents, and change
CA10 AI application capstone and professional handover
- CA10-U01 Scope, users, and acceptance contract Define a useful bounded problem with a manual baseline
- CA10-U02 Integrated build and source controls Connect data, interface, model, and permitted tools
- CA10-U03 Evaluation, usability, and adversarial review Test success, uncertainty, misuse, and recovery
- CA10-U04 Live defense and handover Diagnose an unseen fault and enable peer reproduction
CA20 Data and model capstone and research communication
- CA20-U01 Problem formulation and dataset acceptance Define target, leakage risks, provenance, and permitted use
- CA20-U02 Baselines and reproducible experiment Record code, data, environment, and comparison conditions
- CA20-U03 Model improvement and ablation Test one justified change without contaminating final evaluation
- CA20-U04 Robustness, fairness, and resource tradeoffs Analyze slices and operational consequences
- CA20-U05 Serving or batch workflow and monitoring Demonstrate a bounded scoring and response process
- CA20-U06 Independent defense and handover Explain limitations and reproduce the study on unfamiliar inputs