diff --git a/lava_utils.sh b/lava_utils.sh
new file mode 100644
index 0000000..7040454
--- /dev/null
+++ b/lava_utils.sh
@@ -0,0 +1,255 @@
+#!/usr/bin/env bash
+#
+# Copyright (c) 2019-2022, Arm Limited. All rights reserved.
+#
+# SPDX-License-Identifier: BSD-3-Clause
+#
+
+gen_lava_job_def() {
+	local yaml_template_file="${yaml_template_file:?}"
+	local yaml_file="${yaml_file:?}"
+	local yaml_job_file="${yaml_job_file}"
+
+	# Bash doesn't have array values, we have to create references to the
+	# array of artefacts and their urls.
+	declare -n artefacts="$1"
+	declare -n artefact_urls="$2"
+
+	readarray -t boot_arguments < "${lava_model_params}"
+
+	# Source runtime environment variables now so that they are accessible from
+	# the LAVA job template.
+	local run_root="${archive}/run"
+	local run_env="${run_root}/env"
+
+	if [ -f "${run_env}" ]; then
+	    source "${run_env}"
+	fi
+
+	# Generate the LAVA job definition, minus the test expectations
+	expand_template "${yaml_template_file}" > "${yaml_file}"
+
+	gen_yaml_expect >> "$yaml_file"
+
+	# create job.yaml
+	cp "$yaml_file" "$yaml_job_file"
+
+	# archive both yamls
+	archive_file "$yaml_file"
+	archive_file "$yaml_job_file"
+}
+
+gen_lava_model_params() {
+	local lava_model_params="${lava_model_params:?}"
+	declare -n macros="$1"
+
+	# Derive LAVA model parameters from the non-LAVA ones
+	cp "${archive}/model_params" "${lava_model_params}"
+
+	sed -i '/^[[:space:]]*$/d' $lava_model_params
+
+	if [[ $model =~ "qemu" ]]; then
+		# Strip the model parameters of parameters already specified in the deploy
+		# overlay and job context.
+		sed -i '/-M/d;/kernel/d;/initrd/d;/bios/d;/cpu/d;/^[[:space:]]*$/d' \
+				$lava_model_params
+	elif [[ ! $model =~ "qemu" ]]; then
+		# FIXME find a way to properly match FVP configurations.
+		# Ensure braces in the FVP model parameters are not accidentally
+		# interpreted as LAVA macros.
+		sed -i -e 's/{/{{/g' "${lava_model_params}"
+		sed -i -e 's/}/}}/g' "${lava_model_params}"
+	else
+		echo "Unsupported emulated platform $model."
+	fi
+
+	# LAVA expects binary paths as macros, i.e. `{X}` instead of `x.bin`, so
+	# replace the file paths in our pre-generated model parameters.
+	for regex in "${!macros[@]}"; do
+		sed -i -e "s!${regex}!${macros[${regex}]}!" "${lava_model_params}"
+	done
+}
+
+gen_yaml_template() {
+	local target="${target-fvp}"
+	local yaml_template_file="${yaml_template_file-$workspace/${target}_template.yaml}"
+
+	local payload_type="${payload_type:?}"
+
+	cp "${ci_root}/script/lava-templates/${target}-${payload_type:?}.yaml" \
+		"${yaml_template_file}"
+
+	archive_file "$yaml_template_file"
+}
+
+gen_yaml_expect() {
+	# Loop through all uarts expect files
+	for expect_file in $(find $run_root -name expect); do
+		local uart_number=$(basename "$(dirname ${expect_file})")
+
+		# Only handle the primary UART through LAVA. The remaining UARTs are
+		# validated after LAVA returns by the post-expect script.
+		if [ "${uart_number:?}" != "uart$(get_primary_uart "${archive}")" ]; then
+				continue
+		fi
+
+		# Array containing "interactive" or "monitor" expect strings and populated during run config execution.
+		# Interactive expect scripts are converted into LAVA Interactive Test Actions (see
+		# https://tf.validation.linaro.org/static/docs/v2/interactive.html#writing-tests-interactive) and
+		# monitor expect scripts are converted into LAVA Monitor Test Actions (see
+		# https://validation.linaro.org/static/docs/v2/actions-test.html#monitor)
+		#
+		# Interactive Expect strings have the format 'i;<prompt>;<succeses>;<failures>;<commands>'
+		# where multiple successes or  failures or commands are separated by @
+		#
+		# Monitor Expect strings have the format 'm;<start>;<end>;<patterns>'
+		# where multiple patterns are separated by @
+		#
+		expect_string=()
+
+		# Get the real name of the expect file
+		expect_file=$(cat $expect_file)
+
+		# Source the run_config enviroment variables
+		env=$run_root/$uart_number/env
+		if [ -e $env ]; then
+			source $env
+		fi
+
+		# Get all expect strings
+		expect_dir="${ci_root}/expect-lava"
+		expect_file="${expect_dir}/${expect_file}"
+
+		# Allow the expectations to be provided directly in LAVA's job YAML
+		# format, rather than converting it from a pseudo-Expect Bash script in
+		# the block below.
+		if [ -f "${expect_file/.exp/.yaml}" ]; then
+			pushd "${expect_dir}"
+			expand_template "${expect_file/.exp/.yaml}"
+			popd
+
+			continue
+		else
+			source "${expect_file}"
+		fi
+
+		if [ ${#expect_string[@]} -gt 0 ]; then
+			# expect loop
+			for key in "${!expect_string[@]}"; do
+				# single raw expect string
+				es="${expect_string[${key}]}"
+
+				# action type: either m or i
+				action="$(echo "${es}" | awk -F ';' '{print $1}')"
+
+				if [ "${action}" = "m" ]; then
+					start="$(echo "${es}" | awk -F ';' '{print $2}')"
+					end="$(echo "${es}" | awk -F ';' '{print $3}')"
+					patterns="$(echo "${es}" | awk -F ';' '{print $4}')"
+
+					cat <<-EOF
+					- test:
+					   monitors:
+					   - name: tests
+					     start: '${start}'
+					     end: '${end}'
+					EOF
+
+					# Patterns are separated by '@'
+					OLD_IFS=$IFS; IFS=$'@'
+					for p in ${patterns}; do
+						cat <<-EOF
+						     pattern: '$p'
+						EOF
+					done
+
+					IFS=$OLD_IFS
+					cat <<-EOF
+					     fixupdict:
+					      PASS: pass
+					      FAIL: fail
+					EOF
+				fi # end of monitor action
+
+				if [ "${action}" = "i" ]; then
+					prompts="$(echo "${es}" | awk -F ';' '{print $2}')"
+					success="$(echo "${es}" | awk -F ';' '{print $3}')"
+					failure="$(echo "${es}" | awk -F ';' '{print $4}')"
+					commands="$(echo "${es}" | awk -F ';' '{print $5}')"
+
+					test_name="${uart_number}_${key}"
+
+					cat <<-EOF
+					- test:
+					   interactive:
+					EOF
+					OLD_IFS=$IFS; IFS=$'@'
+
+					name="${test_name}" \
+					commands=$commands \
+						print_interactive_command
+
+					if [[ -n "${success}" ]]; then
+						 message="$success" \
+							print_lava_result_msg
+                                        fi
+
+					if [[ -n "${failure}" ]]; then
+						res="failures" \
+						message="$failure" \
+							print_lava_result_msg
+					fi
+					IFS=$OLD_IFS
+				fi # end of interactive action
+
+			done # end of expect	loop
+		fi
+	done # end of uart loop
+}
+
+
+print_interactive_command(){
+	local name="${name:?}"
+	local prompt=${prompts:-}
+	local commands=${commands:-}
+
+	cat <<-EOF
+	   - name: interactive_${name}
+	     prompts: ['${prompt}']
+	     script:
+	     - name: interactive_command_${name}
+	EOF
+
+	if [ -z "${commands}" ]; then
+		cat <<-EOF
+		       command:
+		EOF
+	else
+		for c in ${commands}; do
+			cat <<-EOF
+			       command: "$c"
+			EOF
+		done
+	fi
+}
+
+print_lava_result_msg() {
+	local res="${res:-successes}"
+	local message=${message:?}
+
+	cat <<-EOF
+	       ${res}:
+	EOF
+
+	for m in ${message}; do
+		cat <<-EOF
+		       - message: '$m'
+		EOF
+
+		if [ $res == "failures" ]; then
+			cat <<-EOF
+		         exception: JobError
+			EOF
+		fi
+	done
+}